Content area
Programming and software engineering are of keen interest to the blind or visually impaired (BVI) community, spurring accessibility enhancements to programming tools that simplify nonvisual code navigation and debugging. Though these enhancements improve the general accessibility of software engineering, they fail to address accessibility of specialized programming domains, such as user interface design, physical computing and data science due to their reliance on visual code outputs. Consequently, these domains have become inaccessible to BVI developers. My dissertation supports the following thesis: Access to visual code outputs is critical for BVI experts to contribute expertise in high-skilled programming work. New interaction techniques, access to data representations, and data-driven studies are critical to make this visual information in widely used programming tools accessible. In this dissertation, I first present details on three efforts I undertook during my PhD: (1) UITap -- a prototype system that demonstrates a set of new interaction techniques that demonstrate approaches to make user interface development accessible, (2) PSST -- a data sonification toolkit that gives BVI developers the tools to access live data generated by sensors used in physical computing, and (3) a large scale analysis to examine the accessibility of computational notebooks ---popular environments used by data scientists. While these three threads of work contribute to the accessibility of three high skilled programming domains -- user interface development, physical computing, and data science, the rapid adoption of AI assistance may be causing a shift in how we program. I conclude with recommendations drawn from BVI developer experiences, to ensure that this paradigm shift in programming remains accessible.