Infrastructuring for Access: Co-Designing Writing Tools with a Dyslexic Academic
Emily Q. Wang, Aron S. Marie · 2026 · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26) · doi:10.1145/3772318.3791104
Summary
Wang and Marie propose Infrastructuring for Access, a design approach that weaves together HCI's infrastructuring theory (Star, Pipek, Ruhleder) with Disability Studies' critique of ableism and with Repair Studies. Unlike Universal Design and Ability-Based Design, which focus on removing barriers, Infrastructuring for Access explicitly treats disabled practitioners' existing workarounds as creative responses worth institutionalising - shifting the design space from barrier removal to 'paving the desire paths' that disabled people have already worn into globally available tools. The approach is demonstrated through an eight-month (31-week) field deployment case study co-authored with Aron, a multilingual dyslexic linguistic anthropology PhD student writing his dissertation. The authors identify three Points of Infrastructuring (PoIs) in commercial spell checkers where Aron's workflow breaks down: specialised language support, constraints of spell-checker algorithms, and text presentation in GUIs. Each PoI had a corresponding ephemeral workaround - copy-pasting into search engines, triangulating four separate queries to spell one word, using citation managers to avoid misspelling author names, and Google-translating Vietnamese terms before inserting them. The team co-designed Jargon Manager, a toolkit built with Google Apps Script, JavaScript, jQuery, and the Google Sheets API. It comprises a Chrome browser extension (Highlight-to-Save, which captures a word plus URL source and user-entered notes, tenses, and tags into a personal Google Sheet dictionary) and a Google Docs add-on (Custom Autocomplete with ten substring-match patterns including begins-with, includes, notes-that-include, and tag/source matching, plus Insert-into-doc and Load-tenses actions). Methods combined semi-structured interviews, co-design sessions, Aron's design diary, asynchronous messaging, and automated interface logs, analysed using constructivist grounded theory.
Key findings
Across the 31-week deployment Aron added 167 words to his dictionary and inserted 112 words into documents, with weekly additions ranging from 1-31 (median 5) and five distinct episodes of active use punctuated by non-use windows tied to job applications, conferences, and health issues. 41% of words were added via the browser extension after Aron encountered spell-checker breakdowns; 59% came from a single co-design session where the team bulk-imported 98 Zotero bibliographic entries (a Python .bibtex-to-CSV conversion Emily wrote) - demonstrating that collaborative infrastructuring can achieve what Aron could not build alone. The Begins-with autocomplete section dominated selection (roughly 65 insertions), followed by tag-substring matches and notes-that-include. Two emergent Points of Infrastructuring appeared that the original design had not anticipated: integrating external language sources (citation managers, bulk import pipelines) and multilingual typing input, where Aron used Jargon Manager to embed Vietnamese terms into English academic prose without the Google Translate triangulation he had previously relied on. Jargon Manager did not eliminate access labor but shifted it: the ephemeral, repeated labor of re-searching every misspelling was converted into persistent dictionary curation that 'grows with you as a writer.' From the case the authors distil six principles of Infrastructuring for Access: foster mutual learning through community engagement; consider both ability-based and non-disability-specific motivations; recognise workarounds as creative responses while acknowledging they create extra labor; incorporate workarounds into redesigns of globally available tools; evaluate success by interdependence and labor alleviation; and anticipate that infrastructuring continues indefinitely as users adapt to new use cases.
Relevance
For accessibility practitioners, the paper is useful in three distinct ways. First, it is a concrete critique of spell checkers and autocorrect as accessibility infrastructure - a reminder that tools embedded deep inside Office and Google Docs are quietly doing accessibility work for or against dyslexic, neurodivergent, multilingual, and domain-specialist writers, and that 'standard English' baked into these tools is a disability and an equity issue. Second, it offers a transferable methodological pattern: long-form, single-participant co-design with a disabled practitioner as co-author, anchored by a design diary and log data, is a credible alternative to broader but shallower user studies when the goal is to understand workflows over time. Third, the six Infrastructuring for Access principles give teams a vocabulary for arguing that shipping a workaround-as-feature is legitimate design work, not scope creep. Limitations are acknowledged: n=1, a sighted dyslexic collaborator whose interface depends on a visual Google Docs sidebar, no generalisation to screen-reader or voice-based dyslexic writers, and no quantitative access-labor metric. The Jargon Manager toolkit itself is tightly coupled to Google Workspace and will not generalise cleanly to LaTeX, Microsoft Word, or offline environments without further infrastructuring work.
Tags: dyslexia · print disability · writing tools · spell checkers · infrastructuring · access labor · co-design · disability studies · assistive technology · neurodiversity · multilingual writing · participatory design · technological ableism