Auditory Navigation in Hyperspace: Design and Evaluation of a Non-Visual Hypermedia System for Blind Users
Sarah Morley, Helen Petrie, Anne-Marie O'Neill, Peter McNally · 1998 · Proceedings of the Third International ACM Conference on Assistive Technologies (Assets '98) · doi:10.1145/274497.274516
Summary
This paper from the University of Hertfordshire's Sensory Disabilities Research Unit presents the design and rigorous evaluation of DAHNI (Demonstrator of the ACCESS media Non-Visual Interface), a hypermedia system built specifically for blind users as part of the EU-funded ACCESS Project. Rather than retrofitting accessibility onto a visual browser, the researchers designed the system from scratch around non-visual navigation principles. DAHNI used a spatial "Workspace" interface shaped like a sideways H, where commands were arranged logically and accessed consistently across three input devices: a standard keyboard, a joystick, and a custom-made touch tablet with tactile overlays. The system addressed fundamental challenges of auditory information processing — that audio is transitory (once heard, it's gone), that users need multiple granularities of navigation (letter, word, sentence, paragraph, whole page), and that orientation cues available visually (font size, layout position) must be conveyed through other means. Non-speech sounds played a central role: earcons and auditory icons provided feedback for commands, confirmed navigation between nodes (a "fanfare" for arriving at a new heading, "paper riffle" for reading commands, "transporter" for jumping home), and identified links (a "bing" sound before the link word spoken in higher pitch). The system was evaluated over several weeks with 9 visually impaired students aged 13-19, using a 37-node educational hypermedia module called "A Tour Around London: Past and Present" that included synthetic speech, digitised audio clips, sound effects, music, and descriptions with tactile pictures.
Key findings
The evaluation used a comprehensive multi-faceted methodology combining objective performance measures (computer logging, video recording, experimenter notes) with subjective ratings, questionnaires, and separate experiments on incidental learning and sound memorability. Key results showed that participants started using the system quickly and efficiently regardless of prior computer experience. The keyboard and touch tablet were preferred input devices (ease of use ratings of 4.66/5.0 each), while the joystick scored lower (3.20/5.0) due to difficulty with precise movements. Non-speech sounds were rated as easy to remember (4.37/5.0) and initially very useful (4.56/5.0); six of nine participants correctly identified 8 or 9 of the 9 sounds. Critically, participants with sound feedback were more secure, less confused, and chose more appropriate commands compared to participants in earlier evaluations without sound feedback. Participants rarely felt lost (orientation rating 4.11/5.0) and reported low concern when they did (2.11/5.0), indicating that blind users need not feel "lost in hyperspace" when adequate navigational support is provided. Incidental learning was strong — participants correctly identified 83.3% of true and false headings from the module. Digitised audio with sound effects received the highest satisfaction rating (5.0/5.0), followed by tactile pictures with descriptions (4.38/5.0), and synthetic speech (3.77/5.0), suggesting multimedia variety is important to sustain engagement.
Relevance
This paper makes two lasting contributions to accessibility practice. First, it demonstrates the value of designing non-visual systems from the ground up rather than adapting visual interfaces — an approach that produced a more effective and learnable system than screen-reading overlays of the era. This "born accessible" principle remains important today when designing for any modality. Second, it provides one of the most thorough evaluation methodologies for non-visual interface research, combining performance metrics, subjective ratings, and memory experiments across multiple sessions. The comprehensive set of design guidelines for auditory hypermedia navigation — covering commands like "Where Am I?", "Overview", variable reading granularity, non-speech sound feedback, and clear heading signposting — directly anticipated features that became standard in DAISY digital talking books and modern screen reader navigation. The finding that non-speech sounds dramatically improved user confidence and efficiency supports the broader use of audio feedback in accessible interfaces beyond just speech output. For content creators, the paper's emphasis on careful heading structure and multimedia variety as aids to orientation remains directly applicable to web content design today.
Tags: auditory interface · blindness · non-speech sounds · hypermedia · nonvisual navigation · user-centred design · evaluation methodology · assistive technology · earcons