pith. sign in

arxiv: 2603.28494 · v2 · pith:SUUWVGULnew · submitted 2026-03-30 · 🧮 math.HO

Tooling for digital accessibility in mathematics: Quickly build compliant course websites that benefit all students

Matthew McMillan , Eli Boyden This is my paper

Pith reviewed 2026-05-21 09:53 UTC · model grok-4.3

classification 🧮 math.HO
keywords digital accessibilityMathMLmathematics educationADA compliancecourse websitesObsidianQuartzstudent performance
0
0 comments X

The pith

A Markdown workflow with Obsidian and Quartz lets math instructors publish screen-reader compliant course sites in minutes and yields large student performance gains.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents a practical toolchain that converts TeX-style input into MathML-enriched HTML websites meeting current digital accessibility standards. This addresses the gap between available web standards and the PDF-heavy workflows most math instructors still use. An empirical comparison across 31 Calculus II sections shows treatment groups outperforming controls, reaching 2.4 standard deviations above the control mean by the final semester. Student experience surveys found no significant difference, suggesting the accessibility features do not harm perceived quality. The authors also supply a taxonomy of prior approaches and a public setup guide to lower adoption barriers.

Core claim

The proposed system, built on Obsidian for content authoring, Quartz for static-site generation with MathML output, Git for collaboration, and Cloudflare Pages for free hosting, enables instructors to produce fully compliant math course websites with roughly one to two hours of initial setup and minutes per update thereafter. Sections taught with these materials outperformed traditional sections, with the performance gap reaching 2.4 standard deviations in the last semester studied. A follow-up phase is proposed to examine adoption obstacles at other institutions.

What carries the argument

The Obsidian-Quartz-Git-Cloudflare toolchain that converts Markdown with TeX syntax into accessible HTML containing MathML.

If this is right

  • Math departments can meet 2024 ADA Title II requirements without abandoning familiar TeX syntax.
  • Student learning outcomes in calculus improve substantially when materials are delivered in screen-reader parsable form.
  • Setup cost is low enough that the workflow can be adopted without dedicated technical staff.
  • A planned second study will identify remaining institutional barriers to wider use.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar toolchains could extend accessibility gains to other notation-heavy fields such as physics or engineering.
  • If performance benefits hold under random assignment, departments might treat compliant site generation as a standard teaching resource rather than an optional extra.
  • Version-controlled Markdown sources make it easier to share and iterate on accessible problem sets across semesters.

Load-bearing premise

The observed performance improvements come from the accessibility workflow itself rather than from the single instructor who taught all treatment sections.

What would settle it

A multi-instructor randomized study in which sections are assigned to the new system or to standard materials would show whether the 2.4-standard-deviation gain persists when instructor identity is controlled.

Figures

Figures reproduced from arXiv: 2603.28494 by Eli Boyden, Matthew McMillan.

Figure 1
Figure 1. Figure 1: System workflow. The instructor interacts only with Obsidian (left). A single “Commit [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Course website: Calculus II home page with links to materials pages arranged by week. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Course website: Sample homework page showing expandable solution callouts. Solution [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: GPA by semester. Blue bars: control sections; red bars: treatment sections. Error [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: A-band and DFW percentages by semester 𝑄5 : “Ease of navigation of course materials was important for supporting my review study for exams.” Respondents selected from: “Strongly agree,” “Agree,” “Neutral,” “Disagree,” “Strongly disagree.” We analyzed responses using two coding schemes. In the collapsed coding, responses are mapped to three categories reflecting direction of agreement: “Strongly agree” and … view at source ↗
Figure 6
Figure 6. Figure 6: Survey responses by question (collapsed coding). [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Semester-by-semester performance for two other instructors who were new to this course [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
read the original abstract

Public universities in the US must now meet digital accessibility (DA) standards under 2024 updates to Title II of the ADA. For math instructors, course materials must be screen-reader parsable, which standard LaTeX-to-PDF workflows cannot achieve. Despite MathML's availability as a web standard for accessible math, instructor adoption of DA-compliant workflows remains very low, creating a gap between available technology and classroom practice. This paper makes three contributions. First, we present a taxonomy of existing approaches to DA-compliant math content, organized by print (PDF) versus web (HTML) output targets, analyzing tradeoffs for instructor adoption. Second, we describe a free workflow using Obsidian (Markdown-based content management), Quartz (static site generator), Git (collaboration and version control), and Cloudflare Pages (free hosting, private source files) that enables math instructors to create and publish DA-compliant course websites with MathML from TeX-based syntax. Setup takes approximately 1-2 hours; thereafter, site updates occur in minutes via a single command. A public setup tutorial is made available. Third, we present an empirical study of student outcomes across 31 sections of Calculus II over 6 semesters. Sections using the proposed system outperformed controls, with the treatment group reaching 2.4 standard deviations above the control mean in the final semester. Although all treatment sections were taught by one instructor, evidence such as acclimation trajectories of other new instructors suggests the system itself meaningfully contributes to performance gains. A student experience survey shows no statistically significant difference between groups, indicating no negative effect on experience. A proposed second study phase will assess barriers to adoption at other institutions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The paper presents a taxonomy of digital accessibility approaches for mathematical content (print vs. web), describes a free Obsidian-Quartz-Git-Cloudflare workflow for producing ADA-compliant course websites with MathML from TeX syntax (setup in 1-2 hours, updates in minutes), and reports an empirical study across 31 Calculus II sections over six semesters in which treatment sections outperformed controls, reaching 2.4 standard deviations above the control mean in the final semester; a student survey found no significant experience difference, and the authors note that all treatment sections were taught by one instructor but cite acclimation trajectories as supporting evidence for the system's contribution.

Significance. If the performance difference can be credibly isolated from instructor effects, the work supplies a low-friction, reproducible tool that directly addresses the gap between MathML standards and instructor practice under the 2024 Title II updates, together with suggestive outcome data that could encourage broader adoption. The public tutorial and emphasis on version-controlled, screen-reader-compatible output are concrete strengths.

major comments (1)
  1. [Empirical study / student outcomes section] In the empirical study (described in the abstract and the section on student outcomes across 31 sections): the headline result that treatment sections reached 2.4 SD above the control mean is load-bearing for the claim that the workflow benefits students, yet the design assigns every treatment section to a single instructor; the manuscript acknowledges the confound and references acclimation trajectories of other new instructors, but supplies neither raw per-section scores, exact control-instructor assignments, nor any regression specification that includes instructor fixed effects or experience covariates, leaving the 2.4 SD figure compatible with a comparable instructor main effect.
minor comments (2)
  1. [Taxonomy section] The taxonomy of existing approaches would be clearer if the categorization criteria (e.g., adoption barriers, output fidelity) were stated explicitly rather than left implicit.
  2. [Throughout] Ensure every table and figure is referenced in the surrounding text and carries a self-contained caption.

Simulated Author's Rebuttal

1 responses · 1 unresolved

We thank the referee for the constructive review and for recognizing the practical strengths of the tooling workflow, the public tutorial, and the emphasis on version-controlled, screen-reader-compatible output. We address the major comment on the empirical study below, proposing targeted revisions to clarify limitations while preserving the suggestive character of the outcome data.

read point-by-point responses

  1. Referee: In the empirical study (described in the abstract and the section on student outcomes across 31 sections): the headline result that treatment sections reached 2.4 SD above the control mean is load-bearing for the claim that the workflow benefits students, yet the design assigns every treatment section to a single instructor; the manuscript acknowledges the confound and references acclimation trajectories of other new instructors, but supplies neither raw per-section scores, exact control-instructor assignments, nor any regression specification that includes instructor fixed effects or experience covariates, leaving the 2.4 SD figure compatible with a comparable instructor main effect.

    Authors: We agree that the single-instructor design for all treatment sections constitutes a genuine limitation for causal attribution and that the 2.4 SD figure could in principle reflect an instructor main effect. The manuscript already flags this confound and invokes acclimation trajectories of other new instructors as counter-evidence. In revision we will expand the student-outcomes section with additional concrete details on those trajectories (e.g., semester-by-semester performance patterns for instructors new to the course), include an explicit statement that no instructor-fixed-effects regression was feasible given the observational design and small number of instructors, and strengthen the language to present the performance difference as suggestive rather than definitive. We cannot, however, release raw per-section scores or exact control-instructor assignments, as these are protected under institutional student-data policies; we will instead add further aggregated, de-identified trend summaries to improve transparency within those constraints. revision: partial

standing simulated objections not resolved
  • Release of raw per-section scores or exact instructor assignments, which are precluded by student privacy regulations.

Circularity Check

0 steps flagged

No significant circularity in workflow description or empirical claims

full rationale

The manuscript describes a taxonomy of DA approaches, details a concrete Obsidian-Quartz-Git-Cloudflare workflow for producing MathML-compliant sites, and reports direct empirical comparisons of student outcomes across 31 sections. None of these steps reduce to self-definition, fitted parameters renamed as predictions, or load-bearing self-citations; the performance difference is presented as an observed mean separation rather than a quantity derived from the system itself by construction. The single-instructor confound is explicitly acknowledged and addressed via external reference to other instructors' trajectories, without circular redefinition of the result. The paper remains self-contained against its stated benchmarks of accessibility compliance and measured section performance.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The paper relies on standard assumptions about ADA compliance requirements and the validity of student performance metrics; no free parameters, new axioms, or invented entities are introduced.

pith-pipeline@v0.9.0 · 5836 in / 1097 out tokens · 53524 ms · 2026-05-21T09:53:53.594955+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

17 extracted references · 17 canonical work pages

  1. [1]

    https://www.ada.gov/resources/2024-03-08-web-rule/, March 2024

    Fact Sheet: New Rule on the Accessibility of Web Content and Mobile Apps Provided by State and Local Governments. https://www.ada.gov/resources/2024-03-08-web-rule/, March 2024

    work page 2024

  2. [2]

    https://www.w3.org/TR/WCAG21/#non-text-content

    Web Content Accessibility Guidelines (WCAG) 2.1. https://www.w3.org/TR/WCAG21/#non-text-content

    work page

  3. [3]

    https://www.w3.org/TR/mathml4/#acc benefits

    Mathematical Markup Language (MathML) Version 4.0. https://www.w3.org/TR/mathml4/#acc benefits

    work page

  4. [4]

    https://ncademi.org/provide/formats/braille-assessments/#c-use-of-mathml-for-interoperability-with-a-variety- of-assistive-technology

    Braille Accessibility in State Assessments: Roles, Responsibilities, and Strategies. https://ncademi.org/provide/formats/braille-assessments/#c-use-of-mathml-for-interoperability-with-a-variety- of-assistive-technology

    work page

  5. [5]

    https://www.freedomscientific.com/training/teachers/accessing-math-content-with-jaws-and-fusion/

    Accessing Math Content with JAWS and Fusion – Freedom Scientific. https://www.freedomscientific.com/training/teachers/accessing-math-content-with-jaws-and-fusion/

    work page

  6. [6]

    Using accessible math textbooks with students who have learning disabilities

    Preston Lewis, Steve Noble, and Neil Soiffer. Using accessible math textbooks with students who have learning disabilities. InProceedings of the 12th International ACM SIGACCESS Conference on Computers and Accessibility, pages 139–146, New York, NY, USA, 2010. ACM. ISBN 978-1-60558-881-0. doi: 10.1145/1878803.1878829

    work page doi:10.1145/1878803.1878829 2010

  7. [7]

    Taylor, John Raible, Jacob Bates, and Jacquelyn J

    Erin Scanlon, Zachary W. Taylor, John Raible, Jacob Bates, and Jacquelyn J. Chini. Physics webpages create barriers to participation for people with disabilities: Five common web accessibility errors and possible solutions.International journal of STEM education, 8(1):1–16, 2021. ISSN 2196-7822. doi: 10.1186/s40594-021-00282-3

    work page doi:10.1186/s40594-021-00282-3 2021

  8. [8]

    Mapping the Landscape of Digital Accessibility in Computer Science Education: A Mapping Literature Review

    Morgan Haley McKie and Alexandra Coso Strong. Mapping the Landscape of Digital Accessibility in Computer Science Education: A Mapping Literature Review. In2024 ASEE Annual Conference & Exposition, June 2024

    work page 2024

  9. [9]

    Jason J. G. White. The Accessibility of Mathematical Notation on the Web and Beyond.Journal of science education for students with disabilities, 23(1):1–14, 2020. ISSN 1940-9923. doi: 10.14448/jsesd.12.0013

    work page doi:10.14448/jsesd.12.0013 2020

  10. [10]

    https://developer.mozilla.org/en-US/docs/Web/MathML, September 2025

    MathML|MDN. https://developer.mozilla.org/en-US/docs/Web/MathML, September 2025

    work page 2025

  11. [11]

    Multimedia platform for mathematics’ interactive learning accessible to blind people.Multimedia tools and applications, 77(5):6191–6208, 2018

    Micha l Ma ´ckowski, Piotr Brzoza, Marek ˙Zabka, and Dominik Spinczyk. Multimedia platform for mathematics’ interactive learning accessible to blind people.Multimedia tools and applications, 77(5):6191–6208, 2018. ISSN 1380-7501. doi: 10.1007/s11042-017-4526-z

    work page doi:10.1007/s11042-017-4526-z 2018

  12. [12]

    Accessible interactive learning of mathematical expressions for school students with visual disabilities.PeerJ

    Amjad Ali, Shah Khusro, and Tahani Jaser Alahmadi. Accessible interactive learning of mathematical expressions for school students with visual disabilities.PeerJ. Computer science, 10:e2599–, 2024. ISSN 2376-5992. doi: 10.7717/peerj-cs.2599

    work page doi:10.7717/peerj-cs.2599 2024

  13. [13]

    MathPlayer v2.1: Web-based math accessibility

    Neil Soiffer. MathPlayer v2.1: Web-based math accessibility. InASSETS ’07 : Proceedings of the Ninth International ACM SIGACCESS Conference on Computers and Accessibility : Tempe, Arizona, USA, October 15-17, 2007, pages 257–258, New York, NY, USA, 2007. ACM. ISBN 978-1-59593-573-1. doi: 10.1145/1296843.1296900

    work page doi:10.1145/1296843.1296900 2007

  14. [14]

    Harvey J. Hindin. MATHTYPE 6.6.Mathematics and computer education, 44(3):273–, 2010. ISSN 0730-8639

    work page 2010

  15. [15]

    EuroMath: A Web-Based Platform for Teaching of Accessible Mathematics

    Donal Fitzpatrick, Azadeh Nazemi, and Grzegorz Terlikowski. EuroMath: A Web-Based Platform for Teaching of Accessible Mathematics. InComputers Helping People with Special Needs, volume 12376 ofLecture Notes in Computer Science, pages 385–392. Springer International Publishing AG, Switzerland, 2020. ISBN 978-3-030-58795-6. doi: 10.1007/978-3-030-58796-3 45

    work page doi:10.1007/978-3-030-58796-3 2020

  16. [16]

    Layout-Aware Text Editing for Efficient Transformation of Academic PDFs to Markdown

    Changxu Duan. Layout-Aware Text Editing for Efficient Transformation of Academic PDFs to Markdown. In Document Analysis and Recognition – ICDAR 2025, Lecture Notes in Computer Science, pages 221–241. Springer Nature Switzerland, Cham, 2026. ISBN 978-3-032-04613-0. doi: 10.1007/978-3-032-04614-7 13

    work page doi:10.1007/978-3-032-04614-7 2025

  17. [17]

    MathML and other XML Technologies for Accessible PDF from LATEX

    Frank Mittelbach, Ulrike Fischer, David Carlisle, and Joseph Wright. MathML and other XML Technologies for Accessible PDF from LATEX. InProceedings of the 2025 ACM Symposium on Document Engineering, pages 1–4, New York, NY, USA, 2025. ACM. ISBN 979-8-4007-1351-4. doi: 10.1145/3704268.3748669

    work page doi:10.1145/3704268.3748669 2025