An Epistemic Injustice Critique of Austin's Ordinary Language Epistemology

Abstract

J. L. Austin argues that ordinary language should be used to identify when it is appropriate or inappropriate to make, accept, or reject knowledge claims. I criticize Austin's account. In our ordinary life, we often accept justifications rooted in racism, sexism, ableism, and classism as reasons to dismiss knowledge claims or challenges, despite the fact such reasons are not good reasons. Austin's Ordinary Language Epistemology (OLE) inadvertently classifies the discounting of knowledge claims in classic cases of epistemic injustice as legitimate ordinary maneuvers. I provide recommendations for revision of OLE and offer a means of distinguishing between dismissals in cases of epistemic injustice and their legitimate counterparts.

Introduction

In this paper I criticize Austin's Ordinary Language Epistemology (OLE) through the lens of social epistemology.¹ In his 1946 paper, “Other minds,” Austin seeks to identify the circumstances under which ordinary language deems it appropriate or inappropriate to make, accept, or reject knowledge claims or challenges to them.² Broadly speaking, OLE suggests that our ordinary practice requires us to have good reasons to claim knowledge, and some special reason to doubt the knowledge claimed by others. And yet, as the epistemic injustice literature of the last decade has made clear, whether or not we accept another's knowledge claims or challenges as legitimate will often be based on ubiquitous social prejudices.

Other feminist philosophers³ have used Austin's concept of speech acts to discuss pornography and free speech, yet none to date has provided a significant analysis of OLE.⁴ Like most views relegated to the history of philosophy, OLE assumes a homogeneous language community dominated by privileged voices. The time is ripe to recast OLE on the basis of a pluralistic language community, which accurately reflects diverse speakers. In this paper, I show that feminist epistemology can provide the resources to challenge and revive OLE.

Below, I argue that fidelity to ordinary language commits us to ordinary justifications for our belief—which are often rooted in racism, sexism, ableism, and classism—as special reasons to dismiss knowledge claims or challenges.⁵ The result is that OLE will classify the discounting of knowledge claims or challenges in classic cases of epistemic injustice as legitimate ordinary maneuvers. Individuals often mistake ordinary justifications as good reasons to dismiss the testimony of others when they are not (though they might falsely believe them to be). In response, OLE must provide a principled means of distinguishing between legitimate dismissals and cases of epistemic injustice.

While OLE is particularly susceptible to this worry, the concern itself is a general one of interest to justice-oriented epistemology. There exists a broader question of how we can show that oppressive background beliefs do not provide us good reasons to doubt the testimony of others.

I begin by showing that the Austinian concept of a special reason to dismiss knowledge claims or challenges is worrisome from the position of epistemic justice. Next, I introduce a number of stipulations, which we can use as criteria to determine which special reasons are good reasons and which result in unjust dismissals. In the final section, I encourage a mutually beneficial exchange of ideas between the defenders of OLE and those in pursuit of epistemic justice.

Austin, testimony, and ordinary language epistemology

In “Other minds,” J. L. Austin argues that what epistemologists say about knowledge is at odds with how “the ordinary man” talks about it in everyday life.⁶ Specifically, Austin has in mind the view of some philosophers that, in order for us to know P, we must rule out any incompatible Q. To know I am human, for instance, I must rule out that I am actually a convincing robot. Yet ordinary practice seems to reject this. We do not need to rule out any possible incompatible Q, only those that are legitimate challenges to our knowledge. In response, Austin argues that in order to determine which challenges to our knowledge claims are legitimate and which are not, we should look to whichever we would find to be legitimate in everyday life (Austin 1956/7, 11).

Mark Kaplan (2018) interprets Austin's OLE as a “fidelity requirement,” which can be understood as a consistency requirement between our ordinary life and our theory of knowledge (Kaplan 2018, 21). The fidelity requirement holds that epistemologists should not be willing to espouse any philosophical doctrine that they would be “unwilling to … act on in ordinary life” (Kaplan 2018, 23). Ordinary language methodology reins in the epistemologist's tendency to concern herself with improbable skeptical scenarios. When one is faced with philosophical statements that contradict our ordinary practice—such as, “We can never know anything about the external world”—one ought to dismiss such statements in favor of ordinary beliefs.

Nancy Bauer (2015) argues that Austin's project is ethical in nature. According to Bauer, Austin's primary insight is that the truth and falsity of language are inextricably bound to question of our responsibility as language speakers.⁷ She writes:

As Austin puts it early in How to Do Things with Words, “our word is our bond.” And to offer our bond is, always, to perform the further illocutionary acts of inviting others to grasp what we mean, risking being misunderstood, and making ourselves vulnerable to the judgments of others … [T]he way we talk is in large part a function of our capacity for taking responsibility for what we do. (Bauer 2015, 91)

Following Bauer, I interpret Austin's philosophy as existing at the intersection of epistemology and ethics. When we dismiss the word of others, we must have good reason to do so; failing this jettisons our responsibility, wronging the speaker.

OLE is an application of ordinary language methodology as it applies to knowledge and testimonial considerations. In his paradigm example of OLE, Austin writes of a goldfinch in the garden:

If you have asked “How do you know it's a goldfinch?” Then I may reply “From its behaviour”, “By its markings,” or, in more detail, “By its red head”, “From its eating thistles”. That is, I indicate, or to some extent set out with some degree of precision, those features of the situation which enable me to recognize it as one to be described in the way I did describe it. Thereupon you may still object …

But that's not enough: plenty of other birds have red heads. What you say doesn't prove it. For all you know it may be a woodpecker …

1. (a) If you say “That's not enough”, then you must have in mind some more or less definite lack. “To be a goldfinch, besides having a red head it must also have the characteristic eye-markings”: or “How do you know it isn't a woodpecker? Woodpeckers have red heads too”. If there is no definite lack, which you are at least prepared to specify on being pressed, then it's silly (outrageous) just to go on saying “That's not enough”.

2. (b) Enough is enough: it doesn't mean everything. Enough means enough to show that (within reason, and for the present intents and purposes) it “can't” be anything else, there is no room for an alternative, competing, description of it. It does not mean, for example, enough to show it isn't a stuffed goldfinch. (Austin 1946/1970, 83–4)

Here, Austin emphasizes two central claims of OLE. First, we give ordinary justifications for our beliefs (“by its red head”), which might be legitimately challenged by a specific reason to think our beliefs are incorrect (“woodpeckers have red heads too”). If there is no specific reason, Austin suggests that we would, in ordinary life, reject the challenge as baseless. Second, the reasons we offer don't have to rule out everything incompatible with them (“a stuffed goldfinch”). We simply need to demonstrate that there is no contextually relevant alternative to what we claim to know.

In sum, we ought to be critical of ordinary standards only when certain cases arise in which there is some specific suggestion of deceit (Austin 1946/1970, 112–13). Austin fails to specify what exactly constitutes a special case, though he does say they are based on observation about language use, and thus, they conform to usage.

I take it that special reasons are those reasons that justify the rejection or dismissal of knowledge claims or challenges they are deployed against. Take, for example, the claim: “I know that I see Barack Obama. He's across the room!” My vision is 20/20, and I'm only 25 feet away. Under ordinary circumstances, it would be acceptable for me to say that “I know that I see Barack Obama.” If, however, I were actually in a wax museum, such a scenario would be a special case—a reason for me to question my knowledge assertion that I am seeing the actual person, Barack Obama, rather than a wax statue of his likeness.

I should note that there is a methodological difference between Austin and myself when it comes to determining the concept of a special reason. Generally speaking, Austin looks at how particular words are used in ordinary conversations in order to attain a philosophical analysis of a concept.⁸ However, we don't often (if ever) use the phrase “special reason” in ordinary conversation. Thus, instead of looking to our ordinary use of this phrase, I consider particular instances where individuals offer special reasons in practice. From these instances, I abstract the concept of a special reason.

The notion of a special reason takes on particular force when it comes to testimonial exchange. Austin's brief treatment of testimony parallels the reasoning for first-hand knowledge claims: we must have a special reason to dismiss or disregard claims made by other speakers (Austin 1946/1970, 82). Whereas in first-hand cases of knowledge, we need only be skeptical of our knowledge claims if there is a reason for us to be suspicious, in second-hand cases of testimony, we have a plethora of special reasons that might undermine knowledge transmission. Austin writes:

It is evident, of course, that this sort of “knowledge” is “liable to be wrong”, owing to the unreliability of human testimony (bias, mistake, lying, exaggeration, &c … Naturally, we are judicious: we don't say we know (at second hand) if there is any special reason to doubt the testimony: but there has to be some reason. It is fundamental in talking (as in other matters) that we are entitled to trust others, except in so far as there is some concrete reason to distrust them. (Austin 1946/1970, 82)

Surely, Austin intends that special reasons must be good reasons. But who judges the quality of such reasons? In the next section, I will argue that the subjective judgments of individual speakers and their language communities result in the perception of certain special reasons as acceptable reasons to doubt testimony even when they are false beliefs.

Special reasons

Austin requires that a legitimate rejection of knowledge claims or challenges must be based on a special reason to believe that belief is incorrect. By ordinary language methodology, two lemmas are at work: (1) concepts must conform to language use, and (2) widespread patterns of usage express semantic rules of use.⁹ Concepts conform to usage in that they are limited to what ordinary language takes them to be—how we actually speak about them.¹⁰ At the same time, our patterns of use give rise to and evolve the concept's meaning.

For example, consider the concept of “emptiness.” By (1), what it means for something to be “empty” will depend on how we use the word in everyday conversation. By (2), if we look at the times in which people use the word “empty,” we can gain an idea of the rule about when it is semantically appropriate to use this concept.

Recall that we are concerned about the concept of a special reason, and that these reasons will be generated from biased sources, which we sometimes rely on in ordinary conversation. From (1), it follows that speakers will take themselves to have the concept of a special reasons to discount knowledge claims or challenges based on their own observations about appropriate use. From (2), it follows that we might track the patterns of usage for what it means to have a special reason to discount or disregard knowledge claims or challenges in order to determine the semantic rules regarding special reasons.¹¹ I will begin by investigating the concept of a special reason by looking at these patterns of use. Next, I will show that OLE—without further stipulation—allows speakers to conflate epistemically unjust special reasons as valid special reasons when they are not.

Though OLE is the featured view of this paper, I take it this problem will generalize for any number of epistemic frameworks that rely on social context as affording reasons to believe or disbelieve.¹² As a result, my solution here will be of interest to the broader question of how we sort legitimate reasons to dismiss from their illegitimate counterparts.

Let us return to our patterns of use, so that we might shed light on what it means to have a special reason to accept or reject knowledge claims or challenges on OLE. One challenge to tracking patterns of use is a trade-off between personal use of terms and use as established by the language community.¹³ On the one hand, the language community will correct an individual's improper language use. Hanfling (1999) writes, “The child who says ‘I swimmed’ instead of ‘I swam’ is taught to mend his language and an adult who made this mistake would not be allowed to get away with it” (Hanfling 1999, 54). In this way, the directionality of ordinary language is top down. And yet, whether or not individuals find something to be appropriate to say is also determined by their own personal reflection. While this may involve considerations of the standards of use within their language community, personal reflection will be colored by an individual's own private beliefs, biases, and personal considerations.

To which patterns do we track in order to conform our concept of a special reason to usage? Those spoken by individual speakers when they take themselves to have special reasons upon personal reflection, or those generally agreed upon by a language community?

Let us distinguish between two patterns of use which, on (1) and (2), ought to determine our concept of a special reason.

Individual Pattern of Use (IPU): Whether a knowledge claim or challenge ought to be accepted, discounted, or ignored is determined by an individual's personal standard for accepting or rejecting the legitimacy of knowledge claims or challenges in their own everyday practice.

Community Pattern of Use (CPU): Whether a knowledge claim or challenge ought to be accepted, discounted, or ignored is determined by a socially or culturally agreed upon¹⁴ notion of what knowledge claims or challenges are acceptable to say or reject at any given time.

One might worry that this dichotomy is too clear cut. After all, we have any number of language communities, which are permeable and fluid in membership.¹⁵ However, the CPU can represent a consensus within any language community. The relationship between IPU and CPU is multifaceted—competent individual speakers give rise to a broader language community, which in turn shapes each individual's use to keep within appropriate community standards.

One way of motivating the rules that guide conversational exchange is to understand language as a “game,” with a “conversational score.” David Lewis (1983) contends that the context of the conversation will include certain presuppositions of certain concepts, tacitly agreed upon by those involved (1983, 393). However, an individual's semantically appropriate use of a concept can vary. When speaker uses the word in a different context, they change the score.¹⁶

Regardless of which pattern we rely on for our concept of a special reason to dismiss knowledge claims or challenges, both IPU and CPU are cause for concern from an epistemic injustice perspective. Epistemic injustice refers to a wrong perpetrated against a knower, either by undermining their ability to know or by deflating the likelihood that their claims will be believed to be knowledge by others.¹⁷ Miranda Fricker (2007) distinguishes two types of epistemic injustice: testimonial injustice and hermeneutical injustice.

Testimonial injustice occurs when prejudice causes a hearer to give a deflated level of credibility to a speaker's word; hermeneutical injustice occurs at a prior stage, when a gap in collective interpretive resources puts someone at an unfair disadvantage when it comes to making sense of their social experiences. (Fricker 2007, 1)

Testimonial injustice occurs between knowers A and B, when A decides, based on unjust (epistemically inadequate) reasons, not to accept B's word that they know that p. In contrast, hermeneutical injustice, far from being an instance, is an ongoing disadvantage—built into our social and cultural fabric—where members of oppressed groups do not have the language to describe or assert knowledge claims regarding a harmful phenomenon.

José Medina (2013) builds on Fricker's account, framing epistemic injustice not only as a moral wrong, but also as a political force requiring epistemic resistance. Epistemic injustice is seen as a consequence of social injustice, which can be resolved insofar as social injustices can be mitigated. Medina writes:

Social injustices typically have a negative impact on our epistemic relations to each other (deteriorating epistemic trust, endangering impartiality, weakening the credibility people ascribe to each other, etc.) and also on our epistemic relations to ourselves (undermining our epistemic confidence, self-trust, and self-reliance; compromising our epistemic goals and projects; weakening our motivation for learning and cognitive improvement, etc.) (Medina 2013, 27)

Epistemic injustice is a serious threat to members of oppressed groups. It limits their ability to transmit knowledge of their experiences to others, and it also prevents individuals from fully recognizing themselves as credible epistemic agents.

The big picture is as follows. On IPU, it would be a sufficient condition for a special reason to dismiss or disregard a knowledge claim or challenge if any speaker had a contextually relevant reason to deem it so. Likewise, on CPU, a sufficient condition for a special reason might be any contextually relevant reason upheld by the dominant language community. Neither of these sufficient conditions will require any relation to justification or truth. Rather, it will be sufficient that either individuals or the dominant language community believe(s) such reasons to be good reasons, whether or not they actually are. As a result, these two patterns will take the social injustices ubiquitous in daily practice and inadvertently condone them as appropriate special reasons to dismiss knowledge claims or challenges. The result will be cases of epistemic injustice, which—on Austin's theory—are appropriate applications of OLE.¹⁸

Having noted the broad-stroke consequences, let us take a more in-depth look.

Community pattern of use

Start with CPU:

Community Pattern of Use (CPU): Whether a knowledge claim or challenge ought to be accepted, discounted, or ignored is determined by a socially or culturally agreed upon¹⁹ notion of what knowledge claims or challenges are acceptable to say or reject at any given time.

If determined by the CPU, the concept of a special reason will rely on a universalized social norm as the basis for judging our knowledge claims. I argue this is deeply problematic because universalized social norms are entrenched in an inequitable hegemonic framework. Such norms often undercut utterances produced by marginalized speakers.

In the United States, a “generalized” social or cultural norm will be perceived as neutral or universal, but will be, in reality, white by default (Fanon 1967). In the Western racial imagination, blackness is perceived as a “heightened negative visibility, whereas whiteness goes unnoticed and becomes a blind spot” (Medina 2013, 215). When we couch ordinary language in terms of this universal it rests on the illusion of a common perspective, which defaults to those identities who hold hegemonic power, minimizing the marginalized standpoint. As a result, tracking the CPU to determine legitimate special reasons would ignore the myriad of identities that define our ordinary practice.

For example, a Black high school student—within his own language community—has a special reason to dismiss his white guidance counselor's knowledge claim that it would be best for him to integrate into the white groups in the cafeteria. However, tracking the CPU (which defaults to a white language community) would discount this student's concerns as an illegitimate special reason. Defaulting to the CPU to establish special reasons fails to take into account the important roles that certain identities play in navigating epistemic values and language games.

When we position ourselves from a place of universality—in this case, some default community pattern from which we assess the appropriateness of knowledge claims and challenges—we embrace a “blindness to difference” (Medina 2013, 151). Thus, by referring to this so-called universal language community to affirm which special reasons are legitimate (at the dismissal of others), we fail to recognize the specificity of the individual experience for people of color, people with disabilities, queers, and women, etc. By binding ourselves to an imagined “neutral” framework, we harmfully eliminate the individual experience of oppressed groups as a factor in assessing epistemic claims.

The CPU also exacerbates hermeneutical injustice. Because the universal community interpretation will default to the white, heterosexual, able-bodied male, there will be an epistemological learning curve for members of oppressed groups, in virtue of their identities. Such individuals will experience a disadvantage in determining which challenges to their knowledge are and are not legitimate.

For instance, if a Black teen expresses knowledge claims about why sitting with other Black kids in the cafeteria is a safer environment for him, and the white guidance counselor challenges this knowledge claim, the teen will be pitted against his own identity-based experiences and the guidance counselor's hegemonic standpoint.²⁰ If the CPU is the pattern that reflects the rules of special reasons, the relevant community for evaluating special reasons will default to the stance of a white, heterosexual, able-bodied male. Thus, marginalized people will need to assimilate their epistemic perspective to this universal framework prior to making epistemic judgments. Members of privileged groups, in contrast, will have fewer or even no such obstacle.²¹

I have argued that universalizing according to community standards of use will result in a racialized neutrality where white is the default in the Western racial imagination. Likewise, I have upheld Medina's notion that blindness to difference will reduce the specificity of other's identities—womanhood, queerness, identification with disability status, or as a person of color—to nothingness in order to share universal commonality in epistemic evaluation.

Now consider the fact that, as a cultural whole, we routinely distrust racial minorities, women, the mentally ill, the poor, among others. When implicit and explicit biases, stereotypes, and assumption of impaired cognitive ability (based on gender, race, or disability status) are woven into our cultural fabric, we can imagine many cases where misogyny, racial prejudice, ableism, or classism might undermine one's trust in believing a knowledge claim posed by another. If we accept that it is often culturally appropriate to distrust individuals for these reasons and we accept that a special reason ought to conform to our pattern of use, it follows that this distrust should be classified as a special reason to discount a knowledge claim or challenge.

Thus, by tracking the CPU and utilizing Austin's account of testimony, we have a special reason to discount the knowledge claims or challenges spoken by members of marginalized groups. Whatever most of the dominant language community deems appropriate, regardless of truth, is a sufficient condition to appropriately dismiss knowledge claims or challenges.

Consider the following case:

A middle school math class is divided up into groups to tackle a set of algebra problems. The group in question is composed of three boys and one girl, Alexa. The boys are stumped by problem 4., but Alexa says, “I know the answer! It's x = 4.” The boys exchange looks, saying, “yeah … okay …” and disregard her knowledge claim, continuing to work on the problems themselves. They don't trust that Alexa knows the right answer—after all, everyone knows girls aren't good at math.

The notion that boys are good at math and girls are good at reading is a socially ingrained, culturally acceptable, false belief.²² By the CPU, the belief that girls have less inherent ability to succeed in math (despite being factually incorrect) would constitute a special reason for the boys to discount Alexa's knowledge claim. By the standard of the boys’ language community, this belief (though, wrong) is a good reason. And yet, the boys’ disregard for Alexa's knowledge claim is a classic case of testimonial injustice.

If our goal is justice-oriented epistemology, I have shown that the CPU is problematic. Is the IPU any better?

Individual pattern of use

Individual Pattern of Use (IPU): Whether a knowledge claim or challenge ought to be accepted, discounted, or ignored is determined by an individual's personal standard for accepting or rejecting the legitimacy of knowledge claims or challenges in their own everyday practice.

Tracking the IPU might seem prima facie implausible—why would anyone think that we should evaluate the concept of a special reason according to an individual's usage, rather than the linguistic community at large? However, basing our concept of a special reason on the usage of the IPU appears to solve many of the problems that arose in the first reading by paying proper respect to personal experiences qua women, qua people of color, qua people with disabilities, and qua queers, etc. in epistemic evaluation.

Still, further examination reveals that this pattern is equally problematic. Ordinary language methodology simply asks us to consider what we would say in various circumstances. It is not meant to filter out linguistic practices that are a result of unjust societal practices, such as biases, stereotypes, or prejudices.²³

Tracking the IPU to determine the concept of a special reason will inadvertently legitimize cases of epistemic injustice regarding self-knowledge as acceptable epistemic results. Imagine the following scenario:

A junior faculty member—Eliza—and a male, senior faculty member—Drew—are discussing a topic about which they are equally qualified. Both have done advanced research on the topic and have published their contributions. During the conversation, Eliza and Drew hit a philosophical disagreement. After several minutes of being told by Drew that she has taken the wrong tack, has failed to consider relevant objections, and that her view is inconsistent, Eliza reduces from knowing that she holds a strong position on the subject to lacking confidence in her original beliefs. Despite the stipulation that Eliza and Drew have equal credentials and quality of evidence for their claims, the combination of Drew's insistence, paired with his membership of a dominant class, is enough for Eliza to second-guess her position.

The asymmetrical power relationship between Eliza and Drew causes a reduction in Eliza's confidence and results in epistemic insecurity. Broadly speaking, members of traditionally oppressed groups—including but not limited to women, people of color, and people with disabilities—are traditionally portrayed as intellectually inferior.²⁴ As a result, “it is not uncommon for members of unjust societies to have distorted images of themselves as knowers: for members of a privileged elite to overestimate their cognitive powers, or for members of underprivileged groups to underestimate their capacities” (Medina 2013, 28). Hence, in this case, an epistemic injustice of self-knowledge.

The IPU, compared with CPU, is even more vulnerable to cases of testimonial injustice, as the prejudices and stereotypes need not be culturally or socially acceptable—they simply need to be held by individual epistemic agents. On this pattern, whether one accepts that another individual has knowledge or that their knowledge claim is legitimate can be undermined by a special reason—where a special reason is any reason (perceived by the speaker to be an acceptable or good reason) not to trust that individual. This will include discounting the knowledge claims and challenges of others in virtue of the identities that the individual is explicitly or implicitly biased against.

Consider this case of testimonial injustice based on the IPU and a special reason backed by explicit bias:

John, an able-bodied man, joins his birdwatching group for a Saturday session. Around noon, John points to a bird and says, “Look, it's a cardinal!” A disabled woman, Suzie, corrects him: “Actually, that has to be a scarlet tanager—I know by the black wings.” John is explicitly biased against disabled people. When he turns to Suzie and sees her physical handicap, he makes the false and insulting assumption that she is mentally impaired as well. As a result of this prejudice, he completely disregards her challenge to his knowledge, checking the cardinal off of his list.

When measured against his individual belief system, John feels that he does have a special reason to cast Suzie's challenge as one to be disregarded, as he believes she is very unlikely to be correct. A perceived cognitive deficiency, though a false assumption grounded by explicit prejudice towards the physically disabled, constitutes a sufficient reason to disregard Suzie's knowledge claim.

While explicit bias results in obvious cases of testimonial injustice on the IPU, cases of implicit biases are also insidious.²⁵

Imagine a commercial flight crew, boarding a plane. The pilot, Aniruddha, is a practicing Sikh who wears the traditional turban. A boy, about to board, asks him, “Do you know how to fly the plane?” He responds, “Not only do I know how to fly this plane, but I also know how to keep everyone safe in case of an emergency.” The boy, reassured, boards with his family. Another hearer in range, Sally, a 65-year-old white woman, is uneasy at the prospect of Aniruddha as her pilot. She does not trust what he has said. She calls the airline, “I just have a bad feeling about this guy. Can you change me to a later flight?”

Here we have a case where the Sally does not have explicit biases against Sikhs, or people she perceives (mistakenly or otherwise) to be Muslim or of Arab descent, she holds an implicit bias towards associating those identities with airplanes due to past terrorist attacks. She does not trust Aniruddha's knowledge claim that he will fly the plane safely, because she has a bad “gut feeling.” By her individual reflection, then, this serves as a legitimate, special reason to distrust the pilot's knowledge claims, and is her impetus to change to a different flight instead.

The IPU is no better than the CPU at avoiding epistemic injustice.

Bad results for OLE

Both the IPU and the CPU allow for the special reasons in cases of testimonial injustice to be taken as an appropriate reason to dismiss, and the epistemic injustice itself is considered an acceptable epistemic result. To review, by lemmas (1) and (2) our concept of a special reason must conform to our usage, and we ought to look at our patterns of usage in order to uncover those concepts. By now, I have made clear that either pattern of use will allow for certain bad (false and unjust) special reasons to be deemed appropriate reasons to dismiss or reject knowledge claims or challenges.

The result is a very thin notion of a special reason, which inadvertently legitimizes cases of epistemic injustice as appropriate dismissals on OLE. On the CPU, special reasons can be based on culturally and socially ingrained prejudices, and on the IPU, special reasons are vulnerable to individual false beliefs about what counts as a special reason. Thus, Austin's account fails to classify cases of epistemic injustice as epistemic injustice because it determines them to be legitimate dismissals of knowledge claims and challenges. Assuming that a legitimate challenge to knowledge claims or challenges would not be categorized as an epistemic injustice, something has gone wrong on OLE.

The first word, not the last

In this section, I investigate how we can salvage OLE from what might appear to be a damning critique. By Austin's own proclamation, “ordinary language is not the last word: in principle it can everywhere be supplemented and improved upon and superseded. Only remember, it is the first word” (Austin 1956/7, 11). Our task is to find a way of properly distinguishing illegitimate special reasons, which cause epistemic injustice, from legitimate special reasons. Once this is in hand, we can properly classify cases of epistemic injustice as epistemic injustice, rather than as acceptable epistemic results. Last, we must diagnose the right kind of reason for why these mistakes have been made. In light of this critique, is OLE salvageable? If so, how?

Let us begin by attempting to use existing features within the view to reconcile this problem. First, could we not simply say that bigots do not qualify as competent speakers? After all, a knowledge claim is supported by ordinary practice only if a competent speaker could reasonably make such a claim in conversation. Perhaps ordinary practice does not support this practice at all.

Unfortunately, even if some of the bigots’ claims will be unreasonable, others might be reasonable (though perhaps morally suspect).²⁶ Further, I take it that being a “competent speaker” requires facility with the semantic and syntactic rules of language. Perpetrating an epistemic injustice does not undercut one's competency as a speaker, only as a moral and epistemic agent.²⁷

Turn to another built-in feature of OLE—correction. I have described that one of the aspects of our language communities is that they influence individual speakers as well as shape appropriate use. Perhaps OLE comes equipped with a self-correcting feature that will rectify these trouble cases: deferring to the CPU keeps the IPU in check.²⁸

I have a few concerns with this approach. First, tracking the CPU will also legitimize special reasons that cause cases of epistemic injustice. So, while this may minimize the problems that occur on the IPU, our notion of a special reason is still vulnerable to producing the patently wrong results. Second, in order for tracking the CPU to prevent cases of testimonial injustice and epistemic injustice on the IPU, it seems to depend, largely, on the willingness of the language community to actually provide correction for inappropriate use. While I do hope that if a person continued to perpetrate epistemic injustices by using overtly racist, sexist, etc. rationales someone would eventually tell them not to say such things, it is not clear that this will be the case. Because certain language communities are insular regarding these prejudices, they will be unable (or unwilling) to self-correct.

My third concern lies with the foggier cases—those of implicit bias, those whose special reasons rely on some facts (but not the relevant facts), and those in which the dismissal is not made in public where correction might take place. In cases of implicit bias, the bias will likely be hidden behind a veneer of legitimacy. Take Sally and her “gut feeling” about Aniruddha. It would be deeply socially inappropriate for a stranger who witnessed this conversation to approach Sally and chastise her. Last, just because an individual commits an epistemic injustice by using inappropriate special reasons to dismiss or disregard knowledge claims or challenges does not mean they will state these reasons in the presence of others or egregiously enough that a correction will be made. For all of these reasons, I do not see correction as a viable solution to the cases that I have raised.

Setting correction aside—could we not just point to the fact that such exchanges are epistemically unjust as a special reason to mark the dismissal as illegitimate? Recall that, according to Austin, we ought to be critical of ordinary standards only when certain cases arise in which there is some specific suggestion of deceit (Austin 1946/1970, 112–13). Cases of epistemic injustice are suspicious—both epistemically and morally. Perhaps this special reason supersedes the illegitimate ones offered by the IPU or CPU.

Unfortunately, this suggestion is rather trivial. It does not offer us a principled means of dividing cases of epistemic injustice from their legitimate counterparts beyond pointing to cases of epistemic injustice and naming them as such.

We must change tack. The problematic feature of OLE is that the language arises from and is cultivated by a culture of use that we know to be prejudiced. It is difficult to remedy such problems from within the very system that yields them. For this reason, let us look not to our socially situated language use, but instead to truth and falsity.

The common feature in these cases of epistemic injustice is that the special reasons are grounded by false beliefs. Let us take the concept of a special reason to be what a legitimate special reason actually is. The problem lies in a gap between the concept of a special reason and the reasons to which the concept is being applied to in practice. When the concept of a special reason is incorrectly applied on the basis of false belief, the dismissers claim to have a legitimate special reason to doubt when they do not.

Systematically false beliefs lead to patterns of usage where, relative to a person's beliefs and/or a community's beliefs, the semantic rules of use are being followed, but the concept is being misapplied. To see this, imagine a southern slave owner. Were we to converse with him, we would likely agree on the semantic rule of what it means to be a “person.” However, when the slave owner is unwilling to assign the term to his Black slaves, we will disagree on the application of this term.²⁹

When individuals and/or language communities hold incorrect beliefs about the world, they take themselves to have good special reasons when such reasons are grounded by false and/or unjustified beliefs. This serves as bad input which produces bad output—it is this misapplication of the concept of a special reason that causes epistemic injustice. Thus, we must find a way to check our patterns of usage to guard against these instances.

To prevent such cases, let us begin by stipulating that, for one to have a special reason to discount a knowledge claim or challenge, this special reason must be a fact about the world. Call this Stipulation 1.

Stipulation 1: For one to legitimately dismiss or discount a knowledge claim or challenge on Austin's OLE, a special reason must be (a) specific, and (b) a fact or facts about the world.

This recommendation seeks to resolve cases of epistemic injustice due to their basis on falsities, which on this view are not candidates for special reasons, since they are not very good reasons at all.

Return to the boys’ dismissal of Alexa. On this view, the Austinian can point out that the boys’ dismissal is grounded by the false belief that girls have less inherent ability to succeed at math problems compared with boys. Because this belief is false, and special reasons are to be facts about the world, they do not have a legitimate reason to dismiss her knowledge claim, and therefore they are participating in an unjust epistemic act.

There is something rather nice about this solution—it captures the failure on the agent's part to give sufficient due to Alexa as an epistemic agent. It also tracks the IPU while remedying the principle's apparent susceptibilities to epistemic injustices. The view can be applied to the other cases of testimonial injustice by claiming that, if the special reason is based on some sexist, racist, classist, or ableist false belief, then such a reason is not a special reason, and therefore cannot be used to dismiss whatever knowledge claim or challenge is being made. The falsity of the belief, then, is precisely what makes the treatment unjust, on the theoretical account.

The view can be extended further to explain epistemic injustice regarding self-knowledge. In such cases, the epistemic agent's deflation results from false beliefs about their own or others’ identities. For instance, in the case of Eliza and Drew, the false belief might be that Drew must know more on the subject, or alternatively, that Eliza must know less. Thus, it is this false belief that results in the unjust deflation of Eliza's confidence in her own knowledge claims.

One might have assumed that these reasons ought to be facts from the get-go, but I have argued that very often our language community and personal self-reflective standards fall short in this regard. Recall that, at most, Austin stated that special reasons must be specific (Austin 1947/70, 84). This particular stipulation has the pleasant corollary of handling certain cases of implicit bias. For instance, the case where Sally “has a bad feeling” about the Sikh pilot is no reason at all, on this view. Austin did not specify, however, whether a special reason should be a fact about the world. Therefore, to use the full force of this remedy, we must stipulate.

My mission has been to shed light on how we can distinguish legitimate special reasons from epistemically unjust reasons masquerading as special reasons. Even beyond any commitment to OLE, one can use this concept of a special reason to provide us a targeted way to delegitimize oppressive beliefs. Legitimate special reasons to dismiss or disregard must be a specific fact about the world, and cases of epistemic injustice will fail one or both criteria.

Still, two cases offered by Ralph DiFranco suggest that Stipulation 1 is insufficient. Take the first case:

We can imagine a hiring manager who gives too much weight to certain objective features of a job candidate in some cases, but gives those same features much less weight in other cases … let's suppose two people apply for a job at a tech company. Candidate A is Black, went to an Ivy League institution, and has ten years of experience in the relevant field. The committee sees these as positive features of A, and so they form a favorable opinion of A. Later on, the committee considers a different candidate, B, who is white, has the same degree from the same institution and the same amount of experience in the field. Yet these features are given more weight when evaluating B, who is seen as an outstanding candidate, and so B is offered the job. Notice that the hiring committee used only facts about the world when assessing A's and B's qualifications and when making the choice to hire B rather than A, yet they didn't assign those facts equal weight in their assessment of each candidate. (DiFranco, personal email, January 7, 2019)

We can make this explicitly a case of epistemic injustice by asserting that both candidate A and B assert during their interview: “I know that I'm the best person for the job.”

DiFranco's case is meant to show that OLE may continue to allow for undesirable epistemic results despite the special reason meeting our factivity requirement. Namely, the committee recognizes the facts about the candidates’ credentials, but weighs them differently. Weighing suggests that one of these beliefs is not necessarily false, but rather that some feature is mistakenly valued greater for candidate B or valued lesser for candidate A.

DiFranco's second case also demonstrates that our current factivity requirement is not quite right:

Imagine a Professor who teaches logic and observes that women students tend not to do well in his class. A woman asks for permission to enroll, and he thinks that she probably won't perform well in the class. The Professor's judgment is determined by facts, but of course there's something deeply wrong with his assessment. Perhaps the reason women don't do well in his course is that he neglects them, doesn't encourage them, and on the first day of class he mentions the fact that women tend not to do well and so their poor performance is due, in part, to stereotype threat. So, the Professor is making an assessment of the knowledge and intellectual ability of prospective women students on the basis of facts only, but he's neglecting relevant facts (viz. they'd do better if he changed his instruction methods). (DiFranco, personal email, 2019)

Again, we might sure up this case as an epistemic injustice by saying that the potential student asserts: “I know I can succeed in this class.” The Professor takes himself to have a special reason to challenge or undercut her knowledge claim that she will do so.

DiFranco's concern is two-pronged. First, he notes the tendency of bigoted people to misuse statistical facts in order to support their own prejudices in their evaluation of other's knowledge and capability. Second, he worries that the sexist Professor fails to take into account all of the relevant information regarding why the women have done poorly in this class.

With this in mind, let us revise the first claim into Stipulation 2:

Stipulation 2: For one to legitimately dismiss or discount a knowledge claim or challenge on Austin's OLE, a special reason must be (a) specific, (b) a fact or facts about the world, and (c) these fact(s) must be at the motivational foundation of the dismisser's judgment to dismiss or disregard the knowledge claim or challenge in question.

By motivational foundation, I mean the fundamental reason(s) on which they ground their epistemic judgment to dismiss or disregard. This will ensure that we are relying on the relevant fact in order to legitimately assess knowledge claims and challenges. I argue that if these justifications are false, sexist, racist, etc. beliefs, then on my account this is precisely what makes them illegitimate ordinary maneuvers.

Return to the case of the hiring discrimination. To get at the motivational foundation for the hiring decision, more information is required. We might ask, “Why did you not give equal weight to the candidate's credentials?” In order for this to remain a case of epistemic injustice (as DiFranco intends), I see two possible answers: the first is implicit bias and the second is explicit bias. If it were implicit bias that caused the hiring committee to improperly compare the qualifications of the two candidates, then this fails Austin's principle that the special reason must be definite—the hiring committee must cite a specific special reason to disregard the knowledge claim. In this case, the hiring committee's dismissal is illegitimate because they are unable to provide an explicit special reason. If they are able to articulate a reason, it will likely rely on explicitly biased or racist beliefs, e.g., “He only got into x ivy league school because he's Black.” In which case, the belief at the motivational foundation of the hiring committee's dismissal is false, and therefore is not a legitimate special reason by our revised stipulation.

Likewise, I believe the case of the sexist logic Professor is also adequately treated by this refinement. The issue in this case is that, to a certain extent, the statistical fact is a surface-level fact. In this context, it seems that the Professor has some sort of obligation to ask why it is the case that women are doing poorly in his class. Thus, to get at the relevant basis for his judgment, we again need additional information. We might ask: “Why do women do poorly in your class?” Our professor might reply “I don't know, they just do” in which case this answer fails Austin's original criteria because he lacks a specific special reason for undermining the knowledge assertion of this new student, unfairly assuming that she will not succeed in the course simply because women have not done well in the past.³⁰ Or he might reply something sexist, such as “Women are worse at logic.” If he does so, then this is a false statement and once again reveals his illegitimate foundation for discouraging the woman from joining the class.

At this point, I have shown that each of the cases of epistemic injustice I have discussed bottoms out in false beliefs. These beliefs serve as the bad input that result in the misjudgment of certain reasons as adequate special reasons when they are not. Perpetrators of epistemic injustice seize on these false beliefs as a special reason to dismiss or disregard a knowledge claim or challenge.

With the introduction of Stipulation 2, I have successfully addressed DiFranco's concern that Stipulation 1 is not strong enough to account for all cases of epistemic injustice in OLE. And yet, there is another concern—that Stipulation 2 is too strong.³¹

If the badness in these cases of epistemic injustice is traced to the falsity of the special reasons at the motivational foundation of the dismisser's judgment, then this classifies the use of justified false beliefs for special reason(s) as equally bad as the use of prejudiced false beliefs.

Take the following case:

As a practical joke, I take Austin to Stuffed Goldfinch National Park, a national park that contains only stuffed goldfinches. I am an expert in ornithology and know that goldfinches migrate south this time of year. Hoping to pull one over on Austin, I fail to tell him the name and nature of the park, and merely suggest we are going for a day of bird watching. As we approach the first display of stuffed goldfinches, I turn to Austin and say, “I bet you $10 you won't see a live goldfinch.” Austin replies, “You're on.” Just then we turn the corner and Austin exclaims “Perfect timing! Look, it's a goldfinch!” Mischievously, I say, “but you don't know it's a live one.” He hands me his binoculars and says, “Of course I know it, here take a look. It's flapping its wings!” To my surprise, a look through the binoculars confirms that Austin has found the only live goldfinch in Stuffed Goldfinch National Park, a chick who has defied migration patterns.

In this case, I have a specific special reason to challenge Austin's knowledge claim that he sees a live goldfinch. The beliefs at the motivational foundation of my dismissal are justified—namely, that we won't see live goldfinches in Stuffed Goldfinch National Park, particularly at this time of year when they have already migrated south. However, these justified beliefs end up being false. So, the falsity alone of these beliefs cannot indicate that something is illegitimate about the knowledge challenge, since my challenge to Austin's knowledge claim was clearly appropriate in these circumstances.

Let us make explicit what has been merely implied—that my account supports an externalist notion of justification. To determine whether a belief is justified, then, we might ask the following question: is the belief determined by a reliable process? Do external facts cause a state of affairs that make the belief likely to be true?

In this way there is a clear distinction between externally justified false beliefs, which ought to be accepted as appropriate special reason(s) to discount claims or challenges, and externally unjustified false beliefs, which ought to be rejected as inappropriate ordinary maneuvers. This specification also allows for the correct interpretation of other case variations of justification and accidental truth.

Consider the case in which someone has a true belief, and yet they still have a special reason to doubt the knowledge claim or challenge. For instance, we can imagine a scenario in which Austin does know the nature of the park, and that it is extremely unlikely that there will be any live goldfinches around this time of year. When he sees the bird flap its wings, he believes, despite special reasons to the contrary, that it is indeed a live bird.

Likewise, we can imagine a case in which Suzie, the physically handicapped bird watcher, is—by coincidence—mentally impaired as well. The accidental truth of this does not vindicate John in this case of epistemic injustice. While he does have a true belief, he still fails to have a special reason to doubt, in the proper deployment of the concept.

Thus, truth is neither necessary nor sufficient. The first example from Stuffed Goldfinch National Park demonstrates that contextually relevant false beliefs can be warranted as a special reason. The modified example of Stuffed Goldfinch National Park shows that a justified true belief can amount to a special reason that is rightfully superseded. Last, the modified example of Suzie and John serves to demonstrate that unjustified true belief makes for an insufficient special reason to dismiss Suzie's knowledge claim.

It is clear now that the common thread is not truth, but justification. Requiring that one be adequately justified³² in believing that their special reason is true will minimize the gap between the set of reasons applied by individuals to the concept of a special reason, and the concept of a special reason itself. We can revise to our final stipulation, Stipulation 3, which will allow for externally justified false beliefs to provide a legitimate basis from which one can dismiss or discount knowledge claims or challenges:

Stipulation 3: For one to legitimately discount or dismiss a knowledge claim or challenge on Austin's OLE, a special reason must be (a) specific, (b) externally justified belief and (c) these beliefs must be at the motivational foundation of the dismisser's judgment to dismiss or disregard the knowledge claim or challenge in question.

When birdwatching in Stuffed Goldfinch National Park during the period where live goldfinches migrate south for the winter, I am externally justified in believing that Austin will not see a live goldfinch. It is terribly unlikely that Austin will be pointing at the only live goldfinch in Stuffed Goldfinch National Park. Thus, the connection between my belief and the truth is quite strong, though—by a twist of fate—this belief turns out to be false. In contrast, Sally is not externally justified in believing that Aniruddha will not fly the plane safely. John is not externally justified in believing that Sally is mentally handicapped. The boys are not adequately externally justified that girls are inherently worse at math. These beliefs are not securely connected to the truth.³³

Conclusion, and upshot

In light of these stipulations, OLE is largely vindicated. When we recognize that a special reason is not grounded on any contextually relevant reason that is perceived to be a good reason by an individual or language community, but rather a reason that is adequately justified in its connection to the truth, we can pinpoint which special reasons are and are not legitimate.

Still, one challenge remains: real-life cases are messy. What do we do in cases where we cannot access or identify if our special reason is externally justified? Assume, for instance, that we take scientific research to be a quality means by which we determine externally justified beliefs about complex, non-perceptual issues. And yet, scientists sometimes disagree about their results.³⁴ For example, there are mixed scientific results about whether there is a connection between one's gender and ability to spatially cognize (Yuan et al. 2019). One might worry, then, that someone could cherry pick their preferred interpretation of the data as a special reason—e.g., “She doesn't know what she's talking about, women can't read maps!” Alternatively, someone could cherry pick the competing interpretation, that “women can read maps as well as anyone.” In cases where the data are mixed, how are we to know whether the special reason to discount (e.g., women have worse special cognition capacities) is externally justified?

In cases such as these, the interlocuter has an epistemic responsibility to check whether there is a scientific consensus for the claims they will be supporting, so as to avoid wrongful claims to external justification based on unrepresentative data. I take it that part of being externally justified is to cohere with (rather than against) scientific consensus.

In the case where there is no scientific consensus at all—as in a true controversy—one simply lacks the external justification required for the specific, special reason to discount. Importantly, justice-oriented epistemology should take a defensive default stance towards knowledge claims, where such claims are presumed competent unless the special reason to disregard is sufficiently justified.

To sum up, this discussion is not only of historical interest. OLE magnifies a timeless injustice—people often take themselves to have good reasons to reject someone's knowledge claims or challenges even when they do not. In this paper I have offered a principled means of rooting out and avoiding unjust or epistemically specious special reasons.

With these added principles, we clear the way to develop a more nuanced understanding of ordinary language and community usage. While it will take time and structural change to afford proper weight to marginalized members of our language community, we can reorient Austin's project away a homogeneous language community and towards a pluralistic one.³⁵

OLE can offer us a positive project against epistemic injustice. We can harness Austin's notion of a special reason and repurpose it so that it not only serves as a defeater, but also as a source of affirmative epistemic action. Namely, we have a special reason to believe during interactions with members of oppressed groups (including ourselves, if we are members of those groups) that we are subject to deflating their claims to knowledge and dismissing their challenges. Interacting with members of oppressed groups, therefore, should give us a specific special reason to reconsider any immediate impulse to dismiss or disregard.³⁶

Ultimately, an exchange of ideas is in order. Just as the recent literature on epistemic injustice has occasioned a revision of the Austinian project, so too can Austin's OLE contribute to the discussion surrounding epistemically unjust beliefs and empowering marginalized testifiers.