Universal Design of Auditory Graphs: A Comparison of Sonification Mappings for Visually Impaired and Sighted Listeners
Bruce N. Walker, Lisa M. Mauney · 2010 · ACM Transactions on Accessible Computing · doi:10.1145/1714458.1714459
Summary
This study investigates whether auditory graphs—using sound to represent data for people who cannot see visual graphs—can be designed universally for both sighted and visually impaired users, or whether different sonification mappings are needed for different populations. Most graphing software is inaccessible to blind users, and while sonification offers an alternative, design decisions about how data maps to sound (polarity, scaling) have historically been based on studies with sighted college students. The researchers used magnitude estimation, a psychophysics method, to determine how 45 blind and visually impaired participants (ages 12-64, recruited from three organizations) interpreted mappings between conceptual data dimensions (temperature, pressure, velocity, size, dollars) and acoustic parameters (frequency/pitch, tempo, modulation index/brightness). Results were compared to data from 435 sighted undergraduates who completed the same procedure. Participants heard sounds varying in frequency (90-3200 Hz), tempo (41-685 bpm), or modulation index (1-10 harmonics) and estimated what value of temperature, size, etc. each sound represented. This revealed both polarity preferences (does higher pitch mean higher or lower temperature?) and scaling functions (how much pitch change represents a given temperature change?). The study also examined whether age of blindness onset affected responses, comparing early-onset (blind before 18) and late-onset participants.
Key findings
Overall, blind and sighted listeners showed remarkably similar preferences. Polarity choices correlated strongly between groups (r = 0.818, p < .001)—when most sighted participants used positive polarity for a mapping, most blind participants did too. Scaling slopes were even more similar (r = 0.963, p < .001), indicating agreement about how much sound change should represent a given data change. However, notable exceptions emerged. For several mappings (frequency:temperature, tempo:size, tempo:dollars, modulation index:size), blind participants showed nearly unanimous positive polarities while sighted participants were split between positive and negative. This means sonification designed for sighted users' "average" preference might use an inappropriate polarity for blind users. The frequency:dollars mapping revealed an intriguing pattern related to blindness onset. Early-onset blind participants preferred negative polarity (higher pitch = fewer dollars), while late-onset participants preferred positive polarity (like sighted users). One blind participant explained the negative preference through real-world experience: a single coin dropped makes a high-pitched clink, while a bag of coins makes a lower thud. Late-onset participants, having experience with visual graphs that typically use positive polarity, may have internalized that convention. The data type being represented matters: the same sound mapped to "pressure" yielded different slopes than when mapped to "temperature," confirming that sonification design must consider both acoustic parameters and semantic meaning.
Relevance
This research provides empirical foundation for designing accessible data sonification tools used in science education and research. The good news for practitioners: universal design is largely achievable—sonification software can often use the same mappings for sighted and blind users. The scaling functions (slopes) are particularly consistent, simplifying implementation. However, designers cannot simply assume what works for sighted users works for blind users. For certain data-to-sound mappings, blind users show more consistent preferences that differ from the split preferences of sighted users. Sonification tools should either match blind users' more unanimous preferences (benefiting both populations) or provide customization options. The finding that age of blindness onset affects interpretation has important implications: "blind users" are not a monolithic group. Users who lost vision later may have internalized visual graphing conventions, while those blind from birth may rely more on real-world auditory experiences. This suggests sonification tools might benefit from user preference settings or brief training to establish expected interpretations. For accessible STEM education, this research validates that auditory graphs can effectively convey data patterns to blind students and scientists, supporting full participation in data analysis activities.
Tags: sonification · auditory display · data visualization · blind users · visual impairment · universal design · psychophysics