Accessible Mathematics: Why Alt Text Is Justthe Beginning

Ask any accessibility specialist working in STEM publishing what keeps them up at night, and the answer is almost always the same: equations.

Alt text has become the baseline response to image accessibility. Add a description, check the box, move on. For photographs, diagrams, and charts, that approach works reasonably well. For mathematics, it falls apart almost immediately.

A screen reader encountering an equation rendered as an image reads out "image of formula" or, at best, a flat text string attempting to describe a quadratic formula in plain English. Mathematical notation is structured, navigable, and spatial. A single description is none of those things.

The solution has existed for decades. The publishing industry has been slow to adopt it at scale.

The Problem with Image-Based Equations

Most equations in STEM publishing today are rendered as images: typeset in LaTeX, exported as a PNG, embedded in the document, and paired with a text alternative. For a sighted reader, nothing is lost. For a reader using a screen reader, almost everything is.

Screen reader software enables blind users to access large segments of electronic content, but this is not true for much of the content written in physics, mathematics, and other STEM disciplines, due to the strong reliance on mathematical symbols and expressions, which screen reader software generally fails to process correctly.

The problem is structural. An image has no internal hierarchy. A fraction has a numerator and a denominator. A matrix has rows and columns. A nested exponent has layers. None of that structure is preserved in a flat image, and none of it can be conveyed by a single text alternative, no matter how carefully written.

Compliance frameworks have begun to catch up with this reality. WCAG standards 1.1 and 1.3.1 cover non-text content and programmatically determinable structure. The National Center on Accessible Digital Educational Materials likewise recommends MathML for its compatibility with screen readers. Modern screen readers such as JAWS and Fusion support MathML parsing with audio and braille output.

What MathML Actually Does

MathML (Mathematical Markup Language) is a W3C standard for encoding mathematical notation in a machine-readable format. Instead of treating an equation as a picture, MathML describes it as a structured tree of elements. A fraction becomes a container with two named children. A square root has an argument. A sum has upper and lower bounds.

Screen readers such as NVDA, JAWS, and VoiceOver can read equations aloud and let users navigate through fractions, subscripts, or matrices element by element. A student can move through a complex equation the way a sighted reader scans it visually: drilling into a denominator, stepping through the terms of a polynomial, checking the limits of an integral.

For braille users, MathML maps directly to established braille math codes including Nemeth and UEB, so a braille display can render equations correctly rather than approximating them in plain text.

The PDF landscape has also shifted significantly. Math in PDF 2.0 is now fully accessible, navigable by assistive technology providing both speech and braille. Foxit Reader and Firefox now support MathML in PDF, and for the first time publishers can produce a single accessible PDF without a separate HTML version for equation accessibility.

Why XML-First Workflows Change the Equation

The challenge with MathML adoption has never been the standard itself. It has been the production workflow.

When a document is built print-first, equations are typeset for appearance and converted to MathML after the fact. At scale, across thousands of journal articles or textbook chapters, that retrofitting is expensive and error-prone.

An XML-first workflow solves this at the source. With JATS, MathML and LaTeX notation can be used together in the same pipeline. When equations are encoded in MathML as part of the primary XML structure, every downstream output inherits that structure automatically. The EPUB gets navigable equations. The HTML gets navigable equations. The PDF, produced to PDF/UA-2 standards, gets navigable equations. Accessibility is not added at the end. It is built in at the beginning.

Universities, journals, and funding agencies increasingly require accessible documents, and the LaTeX Project has been integrating tagging support directly into the LaTeX kernel as of 2025 and 2026. The tools are maturing. The workflow changes required to adopt them are well within reach for most STM publishers.

The Compliance Pressure Is Real

As the European Accessibility Act and the U.S. Americans with Disabilities Act begin imposing new requirements in 2025 and 2026, publishers will need to meet their accessibility obligations with the math in their articles.

The EAA is in force. The ADA Title II deadline for large public institutions is April 2027. Equations rendered as images, with or without alt text, will not satisfy WCAG 2.1 AA requirements for complex mathematical content.

The question is not whether to address mathematical accessibility. It is whether to address it through retrofitting, one title at a time, or through a production workflow that produces accessible output by default.

Where S4Carlisle Fits

S4Carlisle has been working with complex STEM content for over 35 years. Our XML-first and XML-last production workflows support MathML encoding as part of the standard article and book production pipeline, with JATS-compliant outputs that carry equation structure through to HTML, accessible EPUB3, and PDF/UA-2.

For publishers with image-based equation backlists, our team works through remediation systematically, combining automated conversion with specialist human review for complex notation.

Mathematical accessibility is a solved problem. The solution is structured markup, applied at the right point in the workflow.