dialects, charity, and amelioration
(or: verbal and substantive disputes in the metaphysics of gender)
charitability (introduction)
English people and Americans both use the word ‘pants,’ but not in the same way. By ‘pants,’ Americans mean what the British mean by ‘trousers’; the British mean what the Americans mean by ‘underwear.’ At least, these two judgements are one way of interpreting the linguistic data. Another way is to suppose that the British are undergoing a collective national delusion, and really do believe that every pair of underwear is actually a pair of trousers. By ‘pants’ they mean the same thing as the Americans (‘trousers’) and are just systematically confused about the properties of the garments they wear. But this, probably, is not a very good interpretation of the linguistic data. We should probably craft semantic theses about other languages which don’t require us to posit some impossibly pervasive population-level error regarding facts which aren’t so hard to access. That is, our semantic theses should be charitable.
how (not) to be charitable
The sorts of semantic theses constructed in popular conversations about gender tend to be uncharitable. Call a semantic thesis ‘invariantist’ if it assigns one meaning, individually, for each of our gender terms, without permitting any semantic variance based on the dialect of the speaker. A ‘woke’ invariantist semantic thesis might say that gender is determined only by self-identification, and a ‘chudded’ invariantist semantic thesis might say that gender is determined only by one’s assigned sex at birth. These sorts of invariantist theses simply are not going to be charitable: on the woke invariantist thesis, chuds (in classing people who identify as women as men) are not aware of the way that trans people identify; and on the chudded invariantist thesis, wokeists (in classing people assigned male at birth as women) are not aware of the biological features possessed by trans people. But obviously neither of these are the case. Chuds know that some people are trans and identify as a gender that differs from the one they were assigned at birth, and wokeists know that trans people have certain sets of chromosomes or produce certain sets of gametes or whatever else the chud says determines one’s sex. So these invariantist semantic theses require us to say that some large group of English speakers, either wokeists or chuds, don’t know things that they obviously do know. So they’re both wrong.
Note that the problem in both cases is the invariantist aspect of the analysis. Any invariantist analysis of our gender terms is going to end up painting with a brush too broad to account for the variant linguistic data among woke and chudded speakers of English. At least, in so accounting, it will need to posit population-level delusion or stupidity regarding how people identify or their biological characteristics. But this is an undesirable posit to make. So any good descriptive semantic thesis will end up not being invariantist.
Here’s the story I offer in replacement. Just like there are two dialects of English, British and American, which impacts the analytic relations between words like ‘pants,’ ‘underwear,’ and ‘trousers,’ I think there are two dialects of English in respect to our gender terms. Call the first of these Chud Standard English, or CSE, and the second Wokeist Standard English, or WSE. To provide a preliminary distinction between these two dialects: in WSE, ‘trans women are women’ is just true analytically; and in CSE, ‘trans women are women’ is just false analytically. Something like self-id captures WSE’s meanings for our gender terms, and something like a biological analysis captures CSE’s meanings for our gender terms. This sort of story seems to be suggested by the strength of the analogy between (1) British and American variant rulings on propositions containing ‘pants’ and (2) chudded and woke variant rulings on propositions containing ‘woman’ (or ‘man’…). This sort of story also allows us to preserve charity in ways that invariantist rulings do not.
So the dispute between the chud and the wokeist regarding the validity of trans identity is a verbal one. The apparently contradictory claims upheld by either side end up being palatable to both disputants when translated into the same dialect; the dispute is only a consequence of a difference in speaking. To analogize, consider a dispute between an American and an Englishwoman regarding a pair of boxer briefs. The Englishwoman says ‘these are pants,’ the American says, ‘no, they aren’t.’ Superficially, it seems that they disagree. But the appearance of disagreement is really only a consequence of their different dialects. Were we to charitably translate the Englishwoman’s meaning of ‘pants’ into American English, and the American’s meaning of ‘pants’ into British English, they would end up agreeing. That is, if we explain to the American that the Englishwoman is meaning to say that the boxer briefs are undergarments, and to the Englishwoman that the American is meaning to say that the boxer briefs are not undergarments, the dispute would end.
a few sidenotes [27 5 2026]
charity and impropriety
One might argue that there are ways to (1) adopt a chudded invariantist thesis, and (2) account for the patterns of use one sees in putative speakers of WSE without uncharitably positing widespread error on the part of wokeists. One could argue, for example, that these woke instutitions and communities are applying ‘woman’ (and ‘female’, etc.) to trans women for reasons of politeness rather than truth. That is, they refuse to misgender trans women (by calling them ‘male’) because they think that misgendering is impolite and not because trans women really are females. Tomás Bogardus, in his 2026 The Nature of the Sexes, raises a similar point this way:
Glanzberg and Kirk-Giannini may well be confusing social unacceptability or a lack of courtesy for linguistic error, so that what seems true to them is not that such uses of pronouns are linguistically incorrect, but rather that such uses are socially unacceptable (in their social circles, at least). To illustrate this distinction, consider that it would be rude for an undergraduate student to refer to his Ph.D.-holding professor as “Miss Robinson,” or to call her “buddy.” It’s presumptuous, too familiar, uncourteous, etc. But supposing Dr. Robinson is unmarried, and that the undergraduate fully grasps the concept of a buddy, it’s not a linguistic error; it’s not semantically infelicitous. Our judgment that he should use the title ‘Dr.’ and that he should address her in more professional terms would be judgments of social propriety, not a judgment that he’s mistaken about the meanings of these words or the syntax of his constructions.
The idea is that a range of linguistic data (in Bogardus’ case, it’s the intuitions of trans-inclusive speakers, in ours, its their patterns of usage) underdetermines which of two rival theories is true: (1), that people avoid misgendering trans women because it is impolite; or (2), that people avoid misgendering trans women because it is incorrect.
Bogardus’ argument, I think, fails to account for those cases where the socially proper thing to do is to misgender. Imagine this sort of case: one has a transmasculine friend named Theo, and Theo is not yet out to his parents. Theo prefers that, in the company of his family, people refer to him with ‘she,’ ‘her,’ and a deadname instead of his male name and masculine pronouns. In this case, being socially proper is misgendering: it would be rude to subvert Theo’s preferences and possibly out him to his parents. But it seems regardless that the right pronouns for Theo are masculine ones. Speakers of WSE would need to suppress their natural use of ‘he’ and ‘him,’ and, in doing so, would have the feeling that they are lying.
analyticity
It’s natural to ask why the relevant truths (‘trans women are women,’ ‘trans women are not women’) are analytic in WSE and CSE respectively. Amie Thomasson, in her 2024 ‘Should Ontology Be Explanatory?’, offers a good generic answer to this sort of question:
What sort of evidence can we give that a conditional of ordinary English is analytic? One way to go is to appeal to our common responses and linguistic intuitions: We can (with Schiffer) appeal to the felt redundancy of ‘the house is red and the house has the property of redness’. We can appeal to our standard epistemic norms, which don’t require further investigation to move from ‘the house is red’ to ‘the house has the property of redness’, whereas they do require further investigation to move from ‘the house is red’ to ‘the house has a door’. We can (with Strawson and Grice) appeal to the different reactions we might have to someone who said “the house is red and it’s not the case that the house has a door” (doubt or disbelief), versus to someone who said “the house is red and it’s not the case that the house has the property of redness” (bewilderment, not knowing what to make of it). Or we can appeal to our different reactions to someone who says, “the house is red, so the house has the property of redness”, versus “the house is red, so the house has a chimney” (acceptance versus puzzlement at the non-sequitur). We could appeal to the way in which we would correct those who violated or denied this rule—by aiming to understand what they might be trying to say, or attempting to teach them the rule, rather than questioning their evidence. We could appeal to the inappropriateness of inserting ‘probably’ in the “the house is red, so the house probably has the property of redness”, although such insertions are normally appropriate where mere evidential support is in question (“the house is in a snowy region, so it probably has a pitched roof”). And we can note that these are precisely the kinds of clue we use to identify analyticities in other cases—cases metaphysicians have not made contentious. In any case, we can each either do this in the first person, by reflection on our own linguistic intuitions (and on the assumption, for each of us, that we are reasonably typical competent speakers), or we can engage in empirical analysis or experimental work to see how broadly shared these reactions and intuitions are across the linguistic community.
The sorts of data Thomasson mentions in the case of property-talk plausibly carry over, mutatus mutandis, to the case of the categorical gender ascriptions under consideration (replacing Thomasson’s examples concerning property-talk with ones concerning gender-talk is an exercise left for the reader). Of course, keeping in mind Bogardus’ point, the politically laden nature of questions concerning trans womens’ validity generates complications in attempting to carry over Thomasson’s method of demonstrating analyticity in the case of mundane and uncontroversial talk of properties. When Thomasson employs data regarding speaker reactions, for example, (like bewilderment at denying the putatively analytic ‘the house is red, so it has the proprety of redness’), there is minimal worry that the reaction of bewilderment is actually (or underlied by) an expression of political or ethical alignment as opposed to genuine semantic intuitions. This worry, I think, is soothed by attending to the prior response to Bogardus’ point: although affection and politics impact a large portion of cases, there are nonetheless cases (like that of Theo) where these factors are absent or mitigated.
There are two other lines of evidence for the analyticity of the relevant truths. The first is pursued by Cameron Kirk-Gianni in their 2023 ‘Gender First,’ and consists in coming to a ruling on the non-subsectivity of ‘trans’. If ‘trans’ is not a non-subsective adjective, but is instead subsective or intersective, then the inference from ‘x is a trans woman’ to ‘x is a woman’ is valid (and thus, trans women are women, and this is so analytically; the meaning of the adjective guarantees the validity of the inference). The second is the prima facie evidence provided by dictionaries: Merriam-Webster and Cambridge both define ‘trans woman’ such that, as a matter of definition, they are women.
pragmatic questions
Note that this dispute cannot be settled by ‘disproving’ American or British English. These are just languages (or dialects), and languages simply are not candidates for truth values. In our case, calling CSE or WSE ‘false’ is similarly confused. These sorts of statements just involve a category error: only claims are true or false, while languages (and dialects thereof) just are, the way chairs or schools or dances just are. It makes no sense to say that a dance is true, and it makes no sense to say that a dialect is true. They are just different sorts of human practices.
We can, though, condone or sanction different sorts of human practices without calling them true or false. Building schools is good, and this is an activity that I condone and think we should promote, although this activity is not ‘true’. Killing people is bad, and this is an activity that I sanction and think we should eliminate, although this activity is not ‘false’.
The upshot of this observation is that disputes regarding human practices, like killing or building schools or speaking English, are essentially pragmatic. They are, in the first place, questions about what to do and not what is true. The question of whether killing is true can’t even be sensibly raised; only the question of whether killing is conducive to our ends and should be praised or rebuked, promoted or prevented. The same applies to the practice of speaking a language.
To make this point clearer, we can invent some terminology. Factual questions inquire into something’s truth, and pragmatic ones inquire into its efficacy for our ends (whether it is to be done). Internal questions regard statements made within a language (‘internal to’ a language), while external questions regard languages themselves. ‘Is English…?’ is the stem for a type of external question, while ‘Is ‘snow is white’…?’ is the stem for a certain type of internal question. To state the earlier point in this new terminology: internal questions, both pragmatic and factual, have sense, while only one sort of external question has sense, these being pragmatic ones. Factual external questions involve the category error which was pointed out earlier, the attempt to attribute truth to something that just is not a claim or proposition.
So, inasmuch as we are dealing with the question of whether trans women are women, I am skeptical that there are any really factual questions whose resolution would be useful to the dispute. There do, however, remain pragmatic questions about which way of speaking we ought to adopt. That is, neither CSE nor WSE can be ‘disproven,’ but we can elucidate the ways in which either way of speaking is conducive to our ends. The issue is, so to speak, ethical all the way down.
amelioration
Some philosophers think that, for certain purposes, both CSE and WSE are insufficient. For example, philosophers with feminist aims may think that CSE and WSE (that is, self-identification- and biology-oriented meanings for our gender terms) aren’t able to effectively capture the particular ways in which they’d like to speak about gender-based oppression, and so opt to create a novel ‘dialect’ of English which is able to do so more effectively. Sally Haslanger, for example, stipulates definitions of ‘woman’ and ‘man’ which take these properties to be essentially bound up with oppression and privilege, respectively. Haslanger’s definitions aren’t meant to be descriptively adequate (that is, they aren’t meant to really tell us the way that people use words), but are instead meant to serve as replacements for ordinary gender vocabulary in the context of feminist theorizing.
These sorts of ‘ameliorative projects’ have been in vogue in contemporary philosophy of gender. I think that, in the contexts to which they are typically restricted, ameliorative projects really can be successful and can improve the way that feminist discussions are conducted. These projects, in my eyes, are of a piece with the projects undertaken by physicists or mathematicians who choose to adopt natural language terms (like ‘function’ or ‘force’) and give them more precise definitions to be used in the context of the academy or laboratory. However, in the same way that precise, mathematical definitions of ‘function’ are unlikely to take off among laymen (especially not as replacements for the natural language terms), wholly novel ameliorative proposals for our gender terms are unlikely to take off among laymen, too.
So, parallel with contemporary ameliorative projects seeking to propose entirely new ways of speaking (which will remain as specialist jargon), I think we ought to pursue a more pedestrian ameliorative project which seeks to propagate and promote an already-present way of speaking which best serves human aims in general. That is, we should figure out which already-present ways of speaking best promote our goals and choose to focus on supporting and ensuring the proliferation of these ways of speaking.
Think of this project as analogous to a ‘lesser of two evils’ approach to electoral politics. It would be amazing if a party with every correct position were voted in, but every party like this is just too small and novel to have the momentum needed for a successful campaign. Our energy should be placed behind parties which optimize a function of their likelihood to win and the goodness of their policy. In the same way, here, I take it that we ought to promote whichever way of talking about gender that (1) already has a good number of speakers, and (2) best, even if only suboptimally, serves progressive goals.
This requires, I think, an initial descriptive step which conducts an analysis of our gender terms on both CSE and WSE (given that these dialects exhaust the ways in which gender terms are used among English speakers), followed by an evaluative, ameliorative step which tells us which of these dialects best promote human ends.
