A language is, among other things, a systematic pairing of words or sentences with meanings. I will understand this systematic pairing as a function from words to meanings (Lewis 1975). The term ‘language,’ in this paper, refers to this abstract object—a function that takes words as inputs and delivers meanings as outputs.8

Below, I present two views of language-use that disagree about the facts in virtue of which a language user is related to one language function rather than another. Imagine two languages. Language A pairs words with meanings such that the sentence “a cat sat on a mat” is true iff a cat sat on a mat. Language B pairs words with meanings such that “a cat sat on a mat” is true iff a dog slept on a bed. When I use language to speak or interpret speech, I am a speaker or hearer related to Language A rather than Language B.

That said, no individual is ‘related’ to just one language—each language user is related to many languages. A speaker’s relation to one language, call it L1, explains their actual utterance, including any mistakes that they make. A language user is likely related to L1 in virtue of facts about them as an individual (e.g. Chomsky’s I-languages [2006]). Yet another language, call it L2, describes how a speaker ought to speak or how a hearer ought to interpret a given utterance. Famously, facts about a speaker as an individual struggle to capture this sort of normative relation (Kripke 1982). A language user is likely related to L2 in virtue of facts about their linguistic community (Dummett 1993; Devitt 2021; Lepore & Stone 2016).

Of all the languages that we are related to, the one that matters for communicative hermeneutical injustice is the language we use to communicate knowledge. Again, we can use Fricker’s ‘sexual harassment’ case to flesh this out. Language C is English before coinage of the expression sexual harassment. According to Fricker, women once used language C to communicate knowledge to their hearers. Since C had no expression ‘sexual harassment,’ speakers could not communicate knowledge about sexual harassment. Or so Fricker argues.

Views of language-use also sometimes differ on whether I am related to language A or language B. This difference can support or undercut particular views of language-use. In the example above, we have prima facie reason to think that I am related the language where “a cat sat on the mat” is true iff a cat sat on the mat—Language A. So, if an account of language-use entails that I am related to language B, there is prima facie reason to disprefer that account. In §2.3, I use an observation of this sort to undercut Fricker’s preferred account of language, the communitarian view.

On a communitarian view, language users are related to the language used to communicate in virtue of facts about a linguistic community.9 On this view, speaking and interpreting is like borrowing tools from a neighbourhood tool-shed. Further, since our language is determined by facts about our linguistic community, individuals cannot control which tools are in that shed.

Here is a rough version of a popular communitarian view. In my community, language-users expect that “a cat sat on a mat” is used iff it is true that a cat sat on a mat (abstracting away from jokes, hypotheticals, etc.).10 According to this type of communitarian, the pattern of expectations in my community explains why I am related to a language where my utterance of “a cat sat on a mat” is true iff a cat sat on a mat. In other words, the pattern of expectations in my community determines the language-function used to communicate.

On a communitarian view, Non-culpability makes good sense. On this view, the language used to communicate knowledge is always determined by facts about a community. No individual has control over those facts, so no individual has control over the language used to communicate knowledge. Thus, no individual is responsible for hermeneutical gaps in that language, nor for resulting hermeneutical injustices.

Many philosophers have noted that facts about our community are not sufficient to explain language users’ relation to the language used to communicate. Language users often communicate using words, meanings, and word-meaning pairs that differ from their communitarian language—they speak innovatively or, as we may say, introduce an innovative language.11 In what follows, I fold quite different uses of language under “linguistic innovation.” On my usage, an ‘innovation’ is an extremely ordinary part of everyday language use.

A common source of innovation is zero-derivation, where aspects of meaning are preserved but not syntactic type:

In (1), the language used to communicate goes beyond the language user’s communitarian language. Suppose we adopt the communitarian view on which a communitarian language is fixed by the expectations of a linguistic community (the account mentioned above). Community expectations about word order and verb phrases tell hearers that ‘whiskey’ is probably verbified and transitive.13 Community expectations also give us clues to how ‘whiskey’ should be interpreted—it should describe an event that is related to whiskey. But these clues don’t get us very far. In (1), the speaker could mean that she tried to make the punch taste bad, make the punch taste good, or put the punch in a whisky bottle. It is difficult to come up with limits to what the subject did to the punch in (1). If the subject of (1) is well known for drunken conversations with whiskey bottles, then (1) could mean that the subject tried to speak with the punch before the teachers arrived. Suppose that, in the language used to communicate, the speaker’s utterance means “she tried to put whiskey in the punch.” This meaning would not be determined solely by facts about the speaker’s community (Clark & Clark 1979).

Sometimes, speakers abandon lexical similarities to their shared language, preserving only phonetic14 resemblances to their community’s language and using shared expectations about syntax. Coinage of new terms is one such case.15 An English speaker without a word for aeroplanes might describe an aeroplane crash by coining the term ‘thoom’ (imitating the aeroplane’s struggling engine):

In (1) and (2), facts about speakers’ linguistic communities underdetermine the communicated meaning of innovative terms. In (1), a communitarian expression is paired with a new meaning. In (2), a new expression is paired with a new meaning. In these cases, facts about the speakers’ community fail to explain the language users’ relation to the language used to communicate. In other words, the speakers do not use a communitarian language to communicate. They use an innovative language.

These cases, together with the functional definition of a language, give us a neat definition of a linguistic innovation:

It is tempting to dismiss the above toy examples as marginal cases—oddities to thrill philosophical puzzle-solvers. But innovation is crucial to language-use. Communitarian languages are important for speaking with strangers. If I visit Melbourne and ask a stranger, “What’s the time?,” the likelihood that we share a communitarian language also makes it likely that I will get a useful answer. However, communitarian languages are limited—we cannot talk about everything that is important to us. We often need to talk about things our community has never encountered before, or things we regularly encounter that aren’t codified in our community’s language. Perhaps a community of idealised agents with limitless lexical memories, complete knowledge of their world, and endless time to communicate could develop a communitarian language that covered everything they could ever want to talk about. But for agents like us, we need a way to adapt our community’s linguistic tools to whatever context we find ourselves in. (Indeed, a language in which we can share new knowledge is valuable because we are not these idealised agents.)

To make communitarian languages useful, speakers use a wider range of strategies than those exemplified above. They temporarily extend ordinary expressions through metaphor (Sperber & Wilson 1986b), adopt other speakers’ idiosyncrasies (Brennan & Clark 1996), and negotiate the extensions of everything from gradable adjectives (‘tall’) to kind terms (‘athlete’) (Plunkett & Sundell 2013; Ludlow 2014). Communitarian languages also include terms that require innovation. Placeholder names like ‘thingy’ and ‘whatsit’ require language users to assign new meanings on almost every occasion of use. The wide range of context-sensitive expressions identified by contextualists are also good examples of prescribed innovation. Genitives like “John’s book” could be used to talk about the book John owns, the book John wrote, the book John called dibs on, etc. (Recanati 1989: 298). Linguistic innovation is central to human language use, not an annoying, marginal case.17

Throughout the above discussion, I have referred to language users. This includes both speakers and hearers. On my definition of Linguistic Innovation, language users ‘innovate’ if an expression is new to the speaker or hearer. Innovation includes, for example, a teenager who uses well-worn slang that is new to her grandmother. Despite the teenager’s familiarity with the slang term, this counts as a linguistic innovation. The term ‘innovation’ applies awkwardly to such cases. The awkwardness arises because ‘innovation’ is commonly understood as an intentional act performed by an individual. In other words, the term suggests that the innovative expressions are new for the speaker.

In this paper, innovation is a joint action. This is because we are interested in the language used to communicate knowledge. Suppose a speaker suggests an innovation, but the innovation is not taken up by the hearer. In this case, the interlocutors fail to communicate using the innovation, so the innovation is not included in the language used to communicate knowledge. This is an important point for communicative hermeneutical injustice. When an innovation fails, this is a failure of the duo of speaker and hearer. If a member of that duo has failed to do their part, then that member might be culpable for a resulting hermeneutical injustice.

A clearer account of the failure involved in hermeneutical injustice would involve a hyper-local language that a particular speaker-hearer pair cooperatively construct on the fly. Since this involves a duo of speaker and hearer, we might call such a language a “Duolect.”18 There is not space to develop an account of Duolectical language here. For brevity, I will continue awkwardly referring to ‘innovations.’ I briefly return to this point in §4.3.

If a speaker suggests an innovation that is inaccessible to the hearer, the language users will fail to coordinate on an innovation. Suppose I suddenly turn to the person next to me and utter “Fitzbikter,” intending to communicate that the apple I am holding is rotten. In this case, the hearer cannot access the language in which “Fitzbikter” is true iff the apple I am holding is rotten.

Accessibility is a graded notion, with many contributing factors. One factor is the distance between the suggested innovation and the language a hearer brings to the conversation.19 Suppose I intend to describe a restrictive relationship to a monolingual Italian-speaker. I use the metaphorical expression: “He’s suffocating me.” The suggested innovation is so far from the hearer’s prior language (Italian) that it is completely inaccessible to the Italian-speaker. Compare this with uttering the same sentence to a Scots speaker, or an English-speaker from a very different social background, or a close friend. As my hearers get closer to my own linguistic community, their prior languages are better suited to interpreting “He’s suffocating me.”

A speaker also makes an innovation more accessible for their hearer by attending to or providing appropriate clues. Some of these clues already exist in the language users’ context. If I approach the reader of these last few paragraphs and utter “Fitzbikter,” my hearer may already have the clues needed to figure out what I mean. Other clues are provided by the speaker. The speaker may add extra information to the innovation: “Fitzbikter. So you’ll get sick if you eat it.” Or they may explicitly introduce the term: “When I say ‘Fitzbikter,’ I mean ‘This apple is rotten.’ Fitzbikter.” If a speaker suggests an innovation without sufficient clues, the innovation will be inaccessible to the hearer.

A hearer accepts an innovation if they use the innovative pairing of words and meaning to interpret the speaker’s utterance.20 If a hearer does not accept an innovation, then the speaker and hearer fail to coordinate on an innovation. Innovating with language is a perfectly ordinary way of communicating. So, ceteris paribus, a hearer ought to accept a speaker’s innovation. Suppose I point to an apple and say “Fitzbikter” to the reader of this paper. If my hearer dismisses my utterance as nonsense, then the hearer’s contribution to the innovation has failed. (And I would not feel so bad if they ate the apple!)

Here are three candidates for inclusion in the ceteris paribus clause. Firstly, the innovation must be accessible in the sense described above. The hearer does not fail if they do not accept an innovation that they can’t understand. Secondly, interlocutors ought to gain something by innovating. If there is already a usable term in the communitarian language, then the suggested innovation unnecessarily complicates communication. There are many things that a speaker might hope to gain from an innovation. The speaker might aim for secrecy, humour, or they might just want to seem cool. These are legitimate goals for a speaker, and the hearer ought not refuse them. Thirdly, the hearer might be an authority on the subject matter. Experts’ semantic authority ensures that we talk about the same thing when we talk about, e.g., arthritis, bosons, or COVID. An expert usually need not accept a non-expert’s innovation if the innovation involves a presumptively misused technical term.

I have sketched some of the ways in which innovation might fail. Here is a criterion of culpability for hermeneutical injustice based on the above:

If we can get a grip on culpability for innovation failures, then, by the above criterion, we get a grip on culpability for hermeneutical injustice.

We cannot adopt the simple view on which causal responsibility for an innovation failure is culpability for that innovation failure. On such an account, when a marginalised speaker does not provide an accessible innovation, they are a victim of hermeneutical injustice and also culpable for that injustice. So, we need to constrain the account of culpability for innovation failures.

Plausibly, a language user who sincerely contributes to coordinating on an innovation is not culpable if their contribution fails. Speakers who sincerely attempt to communicate are not culpable for innovations that are inaccessible. Similarly, hearers who sincerely attempt to understand an innovation are not culpable if they fail to understand and to accept the innovation.

Individuals might also be excused from culpability for hermeneutical injustice. For example, a hearer might be excused from culpability if they are unluckily distracted when the speaker attempts an innovation. Individuals might also justifiably fail to coordinate on an innovation. Many of the experiences obscured by hermeneutical marginalization are painful to talk about. Sometimes, speakers use obfuscatory language to justifiably hide their experience from others. Hearers may also justifiably refuse to innovate with speakers who suggest morally reprehensible innovations. It is a feature of my view that I can explain these sorts of ‘acceptable hermeneutical injustices’ as justified failures to innovate.

We now have everything needed to assign culpability for a hermeneutical injustice. Here is a case based on Fricker’s “sexual harassment” case.

Kaiya lives in a community without the expression ‘sexual harassment.’ She has been sexually harassed by John. She understands her experience as an experience of unwanted sexual behaviour, so experiences no cognitive hermeneutical injustice. Kaiya speaks to someone in Human Resources about John.

Kaiya uses an innovative metaphor (“John has a thousand hands”) and signals that it will take a bit of effort to grasp the metaphor (“You know what I mean?”). She also provides clues as to how her metaphor should be interpreted. There are many ways that HR could respond to Kaiya. First, consider a response on which HR would be responsible for a resulting hermeneutical injustice:

In this response, HR fails to innovate with Kaiya. Kaiya’s innovation was not inaccessible, so HR is culpable for the failure to communicate and culpable for the resulting hermeneutical injustice.

Second, consider a response on which HR and Kaiya avoid the possible hermeneutical injustice:

In this second response, HR clarifies the metaphor, then accepts Kaiya’s innovation. The process of innovation here is cooperative. Kaiya and HR successfully coordinate on a linguistic innovation that communicates Kaiya’s knowledge of sexual harassment and that avoids a hermeneutical injustice.

When speakers attempt to communicate despite a gap in shared resources, they tend to know what they’re doing. They suggest metaphors, coin terms, and give meaning-clues that a cooperative hearer can pick up on. On Fricker’s view, these speakers are shouting into the wind—a hermeneutical injustice is preordained by gaps in interlocutors’ communitarian languages. On my view, speakers who speak across a gap in shared resources do so rationally, but their hearers often let them down.

Here, again, is the claim I am arguing against:

Fricker assumes that the language used to communicate is always a communitarian language. This leaves out linguistic innovation. When speakers encounter a gap in their community’s language, they can and do innovate over the gap.

Linguistic innovation is a means by which individuals exert control over the language used to communicate knowledge. Speakers exert control over a language by suggesting innovations, hearers exert control by accepting or rejecting innovations.

Taken together, P1 and P2 contradict Non-culpability. Contra Fricker, the language used to communicate is not simply inherited from one’s community. Individuals can exert control over the language used to communicate. If individuals can exert control over the language they use, then they can be culpable for the nature of that language.

The argument so far leaves open a bad response: Perhaps many cases of hermeneutical injustice do not involve linguistic innovation. An objector might constrain the notion of hermeneutical injustice to those cases that do not involve linguistic innovation, arguing that these cases are non-culpable. This is a bad response because communicative hermeneutical injustice necessarily involves attempts at innovation. If a speaker attempts to communicate despite a gap in a language user’s communitarian language, then they attempt to communicate using an expression, meaning, or expression-meaning pair that is not already in one interlocutor’s communitarian resources.21

A testimonial injustice occurs when a hearer assigns an unfairly low credence to a speaker because the hearer employs a prejudiced stereotype when assessing the speaker’s testimony (Fricker 2007: 28, 36). Fricker gives an example drawn from Harper Lee’s (1960) To Kill a Mockingbird. Tom Robinson is a black man accused of raping a white girl, Mayella Ewell. Robinson gives testimony, explaining why he was at Ewell’s house. When Robinson’s jury assesses his testimony, they apply the stereotype of black men as untrustworthy. The jurors assign an unfairly low credence to Robinson’s testimony and convict him—they perpetrate a testimonial injustice against him.

Fricker (2007: 33) argues that testimonial injustice is usually culpable. If my argument is on the right track, then testimonial and hermeneutical injustices are culpable in similar ways.

Stereotypes are inherited from one’s community. Whenever the relevant stereotypes exist in a community, the subjects of these stereotypes are at risk of a credibility deficit—a testimonial injustice. When hearers employ these stereotypes and unfairly evaluate a speaker’s testimony, the hearer perpetrates a testimonial injustice. On Fricker’s account, the hearer is sometimes excused from culpability for this testimonial injustice. A hearer might be so deeply influenced by the prejudices of their day that “it is simply too much to expect the subject to … realign his habits of credibility judgement” (2007: 33).

On my account, hermeneutical injustice operates in a parallel way. When there is a gap in communitarian hermeneutical resources, marginalized individuals are at risk of a communication failure—a hermeneutical injustice. When a hearer employs only these gappy communitarian resources to interpret a speaker’s innovative utterance, the hearer perpetrates a hermeneutical injustice. Hearers are sometimes excused from culpability for this hermeneutical injustice. For example, when an innovation is inaccessible to a hearer, it is simply too much to expect the hearer to accept the suggested innovation.

My account of culpable hermeneutical injustice also evades Fricker’s response to authors like Medina (2012) and Pohlhaus (2011), who articulate culpable forms of motivated ignorance.

Pohlhaus’ wilful hermeneutical ignorance is a motivated refusal to “learn the epistemic resources developed from marginalized groups” (2011: 722). Dominantly situated knowers remain ignorant of marginalized resources because it is in their interests to remain ignorant. For example, some men might have refused to learn the expression ‘sexual harassment’ after it was coined, because it was in their interests to remain ignorant. When such men fail to understand a woman who says, “John sexually harassed me,” they are culpable for their wilful hermeneutical ignorance.

Pohlhaus sees wilful hermeneutical ignorance and hermeneutical injustice as distinct “since the former is not purely structural, but also wilful” (2011: fn 6). Similarly, Mason (2011) and Dotson (2012) articulate culpable epistemic injustices that they take to be distinct from hermeneutical injustice. Mason highlights a kind of unknowing in which members of “dominant groups … fail to have a proper understanding of the social experiences of marginalized groups by disregarding or distorting interpretations offered by marginalized groups” (2011: 306). Dotson introduces “contributory injustice,” which occurs when an agent’s wilful ignorance thwarts a knower’s ability to contribute to shared epistemic resources (2012: 32). These views are distinct, however, as Fricker (2016: 170–176) notes, each author articulates a kind of motivated ignorance.

Medina (2012) finds a closer relationship between hermeneutical injustice and motivated ignorance (specifically, Mills’ [1997] white ignorance). Medina argues that both hermeneutical injustice and motivated ignorance require the same culpable “failures in hermeneutical responsibility” (215). If Medina is right, we can infer from culpability for motivated ignorance to culpability for hermeneutical injustice.

Responding to Medina, Fricker “offer[s] an account of the boundary between” (2016: 170) hermeneutical injustice and the motivated ignorances characterised by Pohlhaus (2011), Mason (2011), Dotson (2012), and Mills (1997). Fricker (170–176) argues that, since hermeneutical injustice and motivated ignorance are distinct, we cannot infer from culpability for motivated ignorance to culpability for hermeneutical injustice.22 Fricker (174–175) accepts that there is overlap between motivated ignorance and hermeneutical injustice. In some “minimal” cases of hermeneutical injustice, the speaker’s community possesses a resource which the hearer’s community does not possess and which the hearer refuses to learn. The resulting communication failure is at once a hermeneutical injustice and motivated ignorance. However, Fricker points out two ways in which hermeneutical injustice and motivated ignorance come apart.

First, many cases of hermeneutical injustice are not cases of motivated ignorance (2016: 172). In some cases of hermeneutical injustice, neither the hearer’s nor the speaker’s community possesses the relevant expression (e.g. the HR case above). However, in cases of motivated ignorance, one community does possess the relevant resources, which another community refuses to learn. Second, many cases of motivated ignorance are not cases of hermeneutical injustice (2007: 176–177). In some cases of motivated ignorance, there is no gap in shared resources. In these cases, the hearer’s community is aware of the relevant epistemic resources, but the hearer refuses to use those resources (e.g. a modern English speaker who treats the expression ‘structural racism’ as nonsensical).

My account of hermeneutical injustice is strikingly similar to the above accounts of motivated ignorance.23 Thus, we might think that my account is also vulnerable to Fricker’s boundary-drawing response. However, my account avoids Fricker’s response in two ways. First, innovation failures do not come apart from communicative hermeneutical injustice. As noted in §3.5, all cases of communicative hermeneutical injustice are failures to innovate across a gap in communitarian resources. If a speaker attempts to communicate despite a situation of hermeneutical marginalization, then she attempts to use an expression, meaning, or expression-meaning pair that is not captured by her communitarian language. She only experiences a communicative hermeneutical injustice if that innovation fails. If an agent does not attempt to communicate across a gap in shared resources (does not attempt to innovate), then there can be no communicative hermeneutical injustice. A communicative hermeneutical injustice occurs only when there is “a more or less doomed attempt on the part of the subject to render an experience intelligible … [to] an interlocutor” (Fricker 2007: 159).

The second way in which my account avoids Fricker’s boundary-drawing response is that innovation failures come apart from motivated ignorance.24

As with hermeneutical injustice, many cases of innovation failure are not cases of motivated ignorance. In many cases of innovation failure, neither the speaker’s nor the hearer’s community possesses the relevant expression. The HR case from §3.4 is of this sort. In these cases, innovation is often thoroughly cooperative—the innovation must be passed back and forth between interlocutors. One interlocutor suggests an innovation, which the other interlocutor then refines. This happens in the good HR case—Kaiya suggests an innovation, which HR refines by clarification. Compare this with cases of motivated ignorance, where the speaker contributes a fully-formed resource that the hearer refuses to use (e.g. a man who refuses to use the expression ‘sexual harassment’).

As with hermeneutical injustice, many cases of motivated ignorance are not cases of innovation failure. Recall our definition of linguistic innovation on which an innovation is an expression, meaning, or expression-meaning pair which does not occur in L, for any L specifiable only with facts about a language user’s community. As Fricker notes (2016: 176–177), in many cases of motivated ignorance, a hearer’s community possesses the relevant epistemic resource, but the hearer refuses to use that resource. In such a case, both the speaker’s and hearer’s communitarian languages already contain the word-meaning pair that the speaker is trying to use. In these cases of motivated ignorance, there is no linguistic innovation, no innovation failure, and no hermeneutical injustice.

Arianna Falbo’s (2022) account of positive hermeneutical injustice highlights a shortcoming in the above discussion of ‘innovation.’25 Falbo argues that Fricker gives an account of negative hermeneutical injustice, which occurs when there is some lack of intelligibility due to a gap in shared resources. In contrast, Falbo’s positive hermeneutical injustice occurs when there is no gap in shared resources, but hermeneutical resources interact to pre-empt the apt use of some resource(s).

Falbo draws an example from the trial of Brock Turner, “a former student-athlete at Stanford University who was discovered raping an unconscious woman” (2022: 349). In letters to the judge presiding over the case, Turner’s friends and family highlighted the ways in which Turner satisfied the concept golden boy—he had many friends and a long list of athletic and academic achievements. Falbo argues that the concepts golden boy and rapist interact to prevent the apt use of the latter concept. rapist applies to a “creepy stranger, loner, monster, or animal” (2022: 349). Falbo argues that some in Turner’s community had difficulty applying rapist to Turner, because they also applied golden boy to Turner. golden boy interacted with rapist to pre-empt the apt use of the latter concept.

As defined above, innovations introduce new resources to deal with hermeneutical gaps. However, in cases of positive hermeneutical injustice, interlocutors do not need a new resource. Rather, they need to change the relationships between extant resources or remove a particular resource. Here is an objection derived from Falbo’s view: individuals can only exert control over the language used to communicate via innovation; individuals cannot control positive hermeneutical injustice via innovation; therefore, individuals have no control over positive hermeneutical injustice and are not culpable for positive hermeneutical injustice.

There are two possible responses to the above objection. First, my argument might only establish culpability for negative hermeneutical injustice. On this response, individuals are culpable for Fricker’s negative hermeneutical injustice, but not for Falbo’s positive hermeneutical injustice.

Second, we can amend the above discussion of ‘innovation’ to support individual culpability for both types of hermeneutical injustice. In §2.4, I noted that a complete account of culpable hermeneutical injustice would use a hyper-local language spoken by a duo of speaker and hearer—their duolect. (Falbo’s point about interacting resources is most compelling for non-linguistic resources. For the purposes of discussion, assume that a duolect includes not just a pairing of words and meanings, but all the epistemic resources needed to communicate knowledge, e.g. concepts, perspectives, etc.) If parts of the language users’ duolect interact to prevent communication, then one interlocutor might experience a positive hermeneutical injustice. However, the nature of this duolect is within interlocutors’ control. Interlocutors introduce innovations into the duolect to deal with negative hermeneutical injustice, or they change the relations between parts of this duolect to deal with Falbo’s positive hermeneutical injustice. Faced with positive hermeneutical injustice, interlocutors in Turner’s community might remove golden boy from their duolect, or they might resist the communitarian clash between golden boy and rapist by simply applying both concepts to Turner. Such interlocutors do not introduce an innovation, but they do speak an innovative duolect.

In §4.2, I noted Fricker’s boundary-drawing response, in which she draws a boundary between motivated ignorance and hermeneutical injustice. I argued that my account avoids Fricker’s response, since all communicative hermeneutical injustices involve a failure to innovate across a gap in shared resources.

However, I can only evade the boundary-drawing response if communicative hermeneutical injustice is a communication failure caused by hermeneutical marginalization. All communication failures caused by hermeneutical marginalization involve a failure to innovate across gappy resources. However, an objector might give an account on which hermeneutical injustice, properly defined, is not a communication failure. I consider a range of such accounts in this sub-section.

The boundary-drawing objector provides an account of hermeneutical injustice that distinguishes hermeneutical injustice from failures to innovate across gappy shared resources. To ease discussion, let’s coin a term for these failures to innovate across gappy resources: “Innovation Injustices.” On my account, hermeneutical injustices and innovation injustices are identical. The objector gives an account on which hermeneutical injustices and innovation injustices are distinct.

In general, a culpable innovation failure is a mere faux pas. When this faux pas occurs against a background of hermeneutical marginalization, we get a hermeneutical injustice. An objector might argue that, since interlocutors are never culpable for this background, they are never culpable for the injustice.

Take the bad case from §3.4, where HR fails to innovate with Kaiya. HR’s rudeness causes an injustice because their rudeness contributes to the hermeneutical marginalization of women. If women were not hermeneutically marginalized, then HR would not cause an injustice. On the current objection, HR is culpable for rudeness, but not for a hermeneutical injustice.

On this objection, I have mischaracterised why hermeneutical injustice is non-culpable.28 On my reading, hermeneutical injustice is non-culpable if individuals lack control over the causal process leading from hermeneutical marginalization to hermeneutical injustice. This takes seriously Fricker’s explanation of why hermeneutical injustice is non-culpable: “Hermeneutical Injustice is a purely structural phenomenon with no individual perpetrator” (2016: 172).

On the objector’s reading, hermeneutical injustice is non-culpable because individuals lack control over the facts in virtue of which an innovation failure is an injustice. Fricker may have this in mind when she says of supposed perpetrators of hermeneutical injustice: “… being culpable for one’s part in a broader injustice makes one a perpetrator only of that part; it does not make one a perpetrator of the broader injustice itself.” (2016: fn 16). Although an individual might be culpable for the failure to innovate, they are not culpable for an injustice, since they could not control whether their failure was a faux pas or a serious injustice.

The objector’s reading of the non-culpability claim opens more issues than I can satisfactorily deal with here. For example, the interlocutor who fails to innovate suffers from bad moral luck—their faux pas, unluckily, turned out to be an injustice (Nelkin 2019; Williams 1976; Nagel 1979). Further, hermeneutical marginalization is built up by a large and diffuse set of individual actions. The hermeneutical injustice, then, can be seen as a collective harm triggered by an interlocutor’s failure to innovate.29 Here is a quick response that suggests a modest conclusion and a strong conclusion for the argument in this paper.

The modest conclusion for this paper is that Non-culpability is false. Stated this way, my argument responds to Fricker’s claim that Hermeneutical Injustice is non-culpable because it is purely structural. The objector’s account of non-culpable hermeneutical injustice does not touch the modest conclusion. The strong conclusion is that there is no plausible account on which hermeneutical injustice is non-culpable. If the objector’s reading of the non-culpability claim is plausible, then my argument does not establish the stronger claim, although my argument does deal with one line of support for the objector’s reading of the non-culpability claim.

We have reason to be hopeful about the stronger conclusion—individuals are often treated as culpable for microaggressions, which have a similar structure to my account of hermeneutical injustice (McTernan 2018; O’Dowd 2018; Friedlander 2018; Rini 2020). A microaggression is a “small act of insult or indignity, relating to a person’s membership in a socially oppressed group, which seems minor on its own but plays a part in significant systemic harm” (Rini 2020: 12). Common microaggressions include racist jokes, interrupting a speaker, and insults. On their own, these acts are wrongful but often insignificant. Against a background of oppression, these microaggressions are a serious wrong. Similarly, on their own, innovation failures are wrongful but often insignificant. Against a background of hermeneutical marginalization, innovation failures are a serious wrong. If individuals are sometimes culpable for microaggressions, we have reason to think that individuals are sometimes culpable for hermeneutical injustice.