← All reviews

How the Alt Text Gets Made: What Roles and Processes of Alt Text Creation Can Teach Us About Inclusive Imagery

Emory James Edwards, Kyle Lewis Polster, Isabel Tuason, Emily Blank, Michael Gilbert, Stacy Branham · 2023 · ACM Transactions on Accessible Computing · doi:10.1145/3587469

Summary

This paper investigates how alternative text is created in industry settings through a collaboration between UC Irvine researchers and Google's Avatar Project, an initiative to create inclusive stickers depicting people with disabilities. The research combines two studies: focus groups and interviews with 25 end users with disabilities about their experiences with alt text, and interviews with 3 artists and 3 accessibility practitioners at Google about their alt text production processes. The researchers identified four distinct alt text creation processes at Google. The User-Evaluation process involves recruiting screen reader users to test whether alt text effectively conveys the intended message. The Lone Writer process places responsibility on individual accessibility advocates who champion alt text quality, often without formal support. The Team Write-A-Thon process brings groups together to rapidly produce alt text for large image sets, emphasizing consistency across descriptions. The Artist-Writer process involves collaboration between the image creator and the alt text author, allowing artistic intent to inform the description. A key contribution is the "layers of interpretation" model, which maps how meaning transforms as images pass through multiple stages: subject identity, commissioner goals, artistic expertise, describer biases, and finally reader interpretation. Each layer adds its own assumptions and potential distortions. The study reveals that artists creating inclusive imagery are themselves interpreting disability identity based on client direction and personal experience, not solely on input from the depicted subjects.

Key findings

The research surfaces three major themes about alt text creation in practice. First, purpose and intentionality matter—image and alt text creators do consider inclusive representation, but the purpose is often defined by organizational clients rather than individual artists. Second, alt text quality is shaped by personal expertise, logistical constraints, and the broader technical systems writers operate within. Third, alt text lacks organizational champions compared to visual content, meaning it often becomes an afterthought in production timelines. End users reported wanting alt text that helps them understand both the literal content and the creator's intent behind images. However, they also expressed concern about how disability is represented in imagery, noting that "inclusive" images can still perpetuate stereotypes if not created thoughtfully. Users valued alt text that acknowledged when images were specifically designed to represent disability experiences. The study found that alt text writers often work without direct feedback from screen reader users, relying on guidelines that may not match actual user needs. The User-Evaluation process was widely preferred by practitioners but rarely implemented due to time and resource constraints.

Relevance

This research has significant implications for organizations creating accessible digital content. It challenges the notion that alt text is simply a technical task of describing images—instead, it's a complex interpretive process that requires dedicated roles, user feedback mechanisms, and integration into UX design workflows. For practitioners, the four process models provide a framework for evaluating current practices and identifying improvements. The finding that alt text lacks organizational advocacy compared to visual content suggests accessibility teams should formalize alt text responsibilities rather than relying on individual champions. The call for user feedback platforms highlights a gap in current tools—organizations need ways to gather screen reader user input at scale. The layers of interpretation model is valuable for training alt text writers to recognize their own biases and consider how descriptions will be received. For inclusive imagery specifically, the research emphasizes that accessibility (having alt text) and inclusivity (respectful disability representation) are distinct but interconnected concerns that both require attention.

Tags: alt text · image accessibility · inclusive imagery · disability representation · organizational practices · UX design · qualitative research