Positionality in the ENABLE Model

A sentence that reads "Captions were not provided" names no one. A sentence that reads "The product team shipped the video without captions, and Deaf viewers relied on auto-generated transcripts that misnamed speakers" names the builder, the decision, the users, and the cost. The two sentences describe the same event. They commit the writer to different positions.

Donna Haraway argued in "Situated Knowledges" (1988) that no analytical account comes from nowhere. Every description of the world carries the reporter's location, interests, and responsibilities. The reporter who pretends otherwise tells the reader less, not more. Haraway named the alternative "situated knowledge": the reporter discloses the angle, the stake, and the limits, and accepts accountability for the partial view the reporter actually has. The ENABLE Model writes from that premise.

What positionality requires in practice

James Charlton's Nothing About Us Without Us (1998) took a disability rights movement slogan and elevated it into a principle of research ethics. The people most affected by an arrangement hold evidentiary authority about it. A researcher who writes about Deaf education without citing Deaf scholars and Deaf-led organizations has produced a weaker account, not a more objective one. The ENABLE Model routes through that principle in two ways. The manifestations corpus documents what disabled people and their organizations have actually done, in their own words where possible and with verbatim quotation when cited. The analytical vocabulary names forces that disabled theorists identified, including Mike Oliver's social model of disability (The Politics of Disablement, 1990), Paul Farmer's structural violence (Pathologies of Power, 2003), Marta Russell's money model of disability (Beyond Ramps, 1998), Alison Kafer's curative imaginary (Feminist, Queer, Crip, 2013), and Sins Invalid's ten principles of disability justice (2015).

The positionality this implies is specific. Lawrence Weru writes the model as a member of the disability community, after years of industry practice across the stages the model now names, and with the academic training described later on this page. The model is not written from outside the arrangement it documents. The builder-side categories were drawn from work on requirement meetings that did and did not name accessibility, design reviews that did and did not include disabled users, development sprints that did and did not ship keyboard operability, QA cycles that did and did not book disabled testers, triage queues that did and did not rank accessibility bugs, iteration cadences that did and did not return to the users who filed the reports, support channels that did and did not stay open, and stopgaps that did and did not turn into permanent architecture. The navigator-side categories were drawn from lived experience with the assistive technologies, third-party tools, system settings, improvised workarounds, human help, feedback channels, legal assertions, and daily endurance that disabled users move through when the upstream work has not been done. When a page describes Nike's FlyEase or AccessiBe's overlay, the page names what third-party journalists, disability organizations, users, and Weru's own practice in both builder and navigator roles have observed. The page does not reproduce the company's marketing language. When Nike's own press materials appear as sources, they appear as claims the company makes about itself, not as neutral description.

The grammar that enforces the position

Grammar carries positionality the way camera angles carry a film's position. Four patterns in English writing about accessibility routinely erase actors, and the ENABLE Model cuts them all.

The linking verb with an abstract subject. "AccessNow is an app that maps accessibility" treats the app as a self-subsisting object. The sentence suppresses the users who built the database, the disabled users who query it, and the accessibility inequities that made the app necessary. Maayan Ziv and her contributors built the AccessNow map because mainstream mapping products did not flag step-free entries or accessible bathrooms. Disabled travelers query the map because they would otherwise arrive at restaurants that turn them away. The active-voice sentence does more analytical work than the stative one.

The passive without an agent. "Accessibility features were added" never names who added them, when, or under what pressure. "Microsoft shipped Narrator in Windows 2000 after the National Federation of the Blind organized a multi-year campaign" names the actor, the artifact, the year, and the pressure that produced the shipping decision. The active-voice sentence is longer, but more useful.

The noun-phrase conversion. "The commitment to inclusion" converts a verb (commit) into a noun (commitment) that hovers in the sentence without requiring anyone to have committed anything. Disability studies calls this move "nominalization," and it routinely allows organizations to claim action they have not taken. "The CEO committed to inclusion in the 2021 annual report and defunded the accessibility team in 2023" names the two actions and lets the reader weigh them against each other.

The category name as grammatical subject. "Disability justice frameworks hold that access is collective" attributes an argument to a category instead of to the people who made the argument. Sins Invalid's ten principles, published in 2015 under the title Skin, Tooth, and Bone, name interdependence as a core commitment. Mia Mingus's 2011 essay "Access Intimacy: The Missing Link" describes access as relational labor. Both of these citations do work that "disability justice frameworks" cannot do. They name the actor, the venue, the year, and the text a reader can consult.

Artifacts have politics, and so do sentences about them

Langdon Winner's 1980 essay "Do Artifacts Have Politics?" argued that technical objects embody political relationships in their design. Winner's central case cited Robert Moses, who built the parkway bridges over Long Island's Southern State Parkway deliberately low so that public buses could not pass underneath, which kept Black and low-income New Yorkers who depended on those buses off the beaches at Jones Island. The bridge is not neutral infrastructure that some groups happen to use less. The bridge is a policy cast in concrete, and it operates twenty-four hours a day without a single additional decision from Moses.

The accessibility artifact operates the same way. A CAPTCHA that offers no audio alternative, a PDF export that strips its tag tree, a form that binds its submit button to a mouseover handler, a video player that plays captions in a font the user cannot resize. Each of these objects makes a decision about who counts as a user, and each one keeps making that decision on every page load. Ruha Benjamin named this pattern "The New Jim Code" in Race After Technology (2019), documenting how default settings, training data, and interface choices reproduce racial hierarchy while wearing the appearance of neutrality. Benjamin's examples include risk-assessment algorithms that encode historical policing patterns and facial-recognition systems trained on skewed datasets. The disability analog is direct. A screen reader that cannot parse a dynamic dropdown, a keyboard trap on a modal, a color contrast ratio below 4.5 to 1, a Google search result whose alt text field sat empty when the engineer shipped. Each artifact concentrates the cost of its political decision on the users the decision excludes.

Winner and Benjamin established that the artifact is already political before the writer reaches for a sentence to describe it. A writer who then describes that artifact in stative, passive, nominalized, or category-subject grammar performs a second political act. The artifact conceals its politics behind the appearance of neutral infrastructure, and the sentence conceals the artifact's politics behind the appearance of neutral description. The two concealments compound. Winner's Moses bridges read as "low bridges" on a highway map the same way "accessibility features were added" reads as a fact report in a quarterly earnings call. Both framings protect the decision-maker from accountability by presenting the decision as if it were a property of the thing rather than the action of a person.

The ENABLE Model's grammar refuses the second concealment, which is the one a writer controls. Naming the builder, the decision, the artifact, and the users affected does not by itself dismantle the artifact. It does make the artifact's politics visible, which is the condition a reader needs before challenging, reversing, or replacing it.

The ENABLE Model is itself an artifact

Winner's argument turns recursively on any account that uses it. If bridges have politics and sentences have politics, then a model that describes bridges and sentences also has politics. The ENABLE Model is not standing outside the arrangement it describes. It is another built thing, with a frontmatter schema, a taxonomy of ten builder-side interventions and nine navigator-side compensations, a set of forces that disable and enable, a manifestations corpus, a writing rulebook, a website hosted on GitHub Pages, and a domain name. Each of those choices embeds a decision about what counts as evidence, what counts as a unit of analysis, and what counts as a valid claim.

The taxonomy cuts the world in a specific way. The ten builder-side categories draw a boundary around what counts as upstream care, and a practice that sits between two categories, for example ongoing maintenance of a legacy screen-reader driver, has to be assigned to one of them or split across them. The nine navigator-side categories do the same on the other side. The model's frontier paragraph requires every manifestation to close by naming what advances the state of the possible, which commits the analysis to a progress vocabulary a reader may or may not accept. The ENABLE acronym itself names "Early Neglect Allows Barriers" and "End-users Navigate Asymmetrical Barriers," which frames accessibility through neglect and labor rather than through, for example, an identity-first or rights-first vocabulary another tradition might prefer. Community membership and industry practice, disclosed earlier on this page, are two of the three sources that shape the model. Academic training is the third, and it cuts the analysis in directions a reader should know about. Lawrence Weru built the model after training that crossed several Harvard departments and a cross-registration at MIT. The capstone landed in Harvard Medical School's Department of Global Health and Social Medicine, which carries Paul Farmer's structural violence and the resocializing disciplines through the analysis. Weru also worked as a researcher at Harvard Medical School's Department of Biomedical Informatics, which shapes the model's attention to systems, data, and the artifacts that carry both. He took undergraduate courses in Harvard's Anthropology department, including Harvard's only anthropology course on disability and the body, and a disability law and policy course at the Harvard Kennedy School. MIT cross-registration added the STS tradition that reads technology as politics, which is why Winner and Benjamin sit comfortably alongside Mike Oliver, Marta Russell, Alison Kafer, Mia Mingus, and Sins Invalid in the citation pattern. The model's analytical lineage draws from all of those trainings. Another reader with a different training, for example a graduate of a deaf studies program, a disability-led cooperative, or a community organizing school, would write a different model. The site runs in English, on GitHub Pages, with footnote-driven citations and a Docusaurus-generated sidebar. Each of those choices impacts who can and can't consume this material.

The manifestations corpus takes evidence from wherever the arrangement was actually documented. Mainstream journalism, specialist publications, disability-led blogs and newsletters, self-advocacy podcasts, oral history interviews, community forums and Discord archives, court records, government filings, and academic papers all qualify as third-party sources when they document what actors did. Community-documented evidence is often the only evidence that reached the arrangement at all, and the model treats it accordingly. The reflexive limit the corpus still carries is reach. A contributor has to find the evidence and write the page, and arrangements that no contributor has yet documented stay invisible to the corpus until someone does that work.

The reflexive move Winner demands is to read the model the same way the model reads the artifacts it catalogs. Which decisions did its builder make, in whose interest, at whose cost? Who does the taxonomy exclude? What evidentiary standards concentrate authority in certain kinds of sources? Whose analytical traditions does the model lean on, and whose does it treat as adjacent rather than constitutive? The positionality post exists to invite that reading rather than to preempt it. A model about artifacts that refused to examine itself as an artifact would reproduce exactly the concealment Winner and Benjamin named. Readers who find the model useful should also find its choices contestable. Readers who find a category, a frontier requirement, or an evidentiary standard wrong for their case have identified a limit of this artifact, and those limits are where the next version of the model, or the next model entirely, starts.

Why the grammar matters to the politics

A writer who lets organizations describe themselves through linking verbs produces promotional prose. A writer who lets passive constructions hide actors produces compliance prose. A writer who converts verbs into nouns produces management consulting prose. None of those registers suit an ethnography of accessibility, because all of them protect the actors whose decisions the ethnography must name.

The active-voice, named-actor grammar the ENABLE Model uses accomplishes three things at once. It discloses Weru's position as the reporter who observed this arrangement. It preserves the agency of disabled users who are navigating, compensating, organizing, and enduring. And it forces the question that an ethnography of care should always force. Who decided, in whose interest, at whose cost?

What readers can do with the position

The ENABLE Model's grammar is not a style preference. It is the minimum the work can do to honor the positionality it claims. Readers who use the model in their own writing, whether on a blog, in a legal brief, in a design review, or in a policy memo, can adopt the same grammar and produce the same analytical clarity. A complaint letter that names the specific product manager, the specific decision, the specific date, and the specific users affected travels further than one that invokes "the lack of accessibility." A design review that names the teammates who flagged the issue, the sprint in which the decision was made, and the disabled user who will switch to an alternative if the fix ships late produces a better conversation than one that says "we need to prioritize accessibility."

The grammar is the politics, in compressed form. The ENABLE Model writes it that way because accessibility work that does not know where it stands cannot discern what it is seeing.