Inspiring Blind High School Students to Pursue Computer Science with Instant Messaging Chatbots
Jeffrey P. Bigham, Maxwell B. Aller, Jeremy T. Brudvik, Jessica O. Leung, Lindsay A. Yazzolino, Richard E. Ladner · 2008 · Proceedings of the 39th SIGCSE Technical Symposium on Computer Science Education (SIGCSE '08) · doi:10.1145/1352135.1352287
Summary
This paper describes the design and delivery of a four-day computer science workshop for fifteen blind high school students at the National Federation of the Blind Youth Slam, where two hundred blind and low vision students explored disciplines often perceived as inaccessible to them. The workshop centered on building instant messaging chatbots — software programs that carry on conversations via IM platforms like Windows Live Messenger. Chatbots were chosen because they are text-based (inherently accessible to screen readers), socially engaging for teenagers, and illustrative of core computer science concepts including artificial intelligence, natural language processing, and web services. Students programmed in C# using a provided framework that abstracted away complexity, requiring only a single method override to create a functional chatbot. The curriculum followed a "programming from the start" philosophy inspired by the Game of Life Workshop for disabled students, putting students into real code immediately rather than using simplified visual environments. The workshop was developed in close consultation with two blind computer users and the NFB. All software was freely available, and an accessible online tutorial used proper heading structure, list elements, and descriptive link text to support screen reader navigation. Each computer was equipped with both a screen reader (JAWS) and screen magnifier (ZoomText), plus audio splitters so instructors could hear what students were hearing.
Key findings
All fifteen students successfully created personalized chatbots despite having no prior programming experience. Students created diverse bots reflecting three themes: technology (retrieving weather, news, and dictionary definitions from web services), dialog (conversational abilities in specific domains), and personality (reflecting their creators' interests and humor). One student's chatbot fooled a friend for eight messages into thinking it was the student. The biggest accessibility barrier encountered was CAPTCHAs — none of the blind students could solve the audio CAPTCHA required to register an IM account, forcing sighted instructors to solve the visual version instead. Coding difficulties unique to blind programmers emerged: screen readers read code serially, making it hard to understand code structure, leading students to type statements outside brackets or inside string literals. A successful coping technique was rapidly moving the cursor through code to build a mental model of the surrounding structure — analogous to how sighted programmers visually scan their code. Six of the fifteen students preferred using residual vision over a screen reader, and several did not know braille, highlighting the diversity within the blind population. Braille code printouts helped students appreciate code structure that was not apparent through audio alone. The 4:1 student-to-instructor ratio proved effective, and blind mentors were invaluable for helping students improve screen reader proficiency even when the mentors lacked CS knowledge.
Relevance
This paper offers practical, transferable guidance for anyone designing accessible computing education. The three design principles distilled — prepare resources in convertible electronic formats, balance technical and accessibility requirements when choosing tools, and test early with the target population — remain foundational for inclusive CS education. The CAPTCHA finding is a vivid example of how a single inaccessible component can block an entire workflow, a lesson still relevant as authentication barriers persist. The observations about code structure comprehension through screen readers — that code's visual layout carries structural meaning invisible to serial audio rendering — anticipated ongoing challenges in making IDEs and programming tools accessible. The workshop demonstrates that blind students can learn real programming with appropriate scaffolding, challenging assumptions that still discourage blind students from pursuing CS. The "programming from the start" approach, using real languages rather than visual block-based tools, is particularly important because most visual programming environments remain inaccessible to screen reader users.
Tags: accessible education · blind students · screen readers · computer science education · inclusive design · CAPTCHA · assistive technology · programming education · universal design