An ethnographic study of visual impairments for voice user interface design

Design for Voice User Interfaces (VUIs) has become more relevant in recent years due to the enormous advances of speech technologies and their growing presence in our everyday lives. Although modern VUIs still present interaction issues, reports indicate they are being adopted by people with different disabilities and having a positive impact. For the first author’s PhD research project, an ethnographic study is currently being carried out in a local charity that provides support and services to people with visual impairments. The purpose is to understand people’s competencies and practices, and how these are, or could be, related to voice technologies (assistive technology and mainstream VUIs). Through direct observation and contextual interviews, we aim to investigate the problems and solutions they encounter and the ways they cope with particular situations.

💡 Research Summary

**
This paper reports on an ongoing ethnographic study that investigates how people with visual impairments use and experience voice‑based technologies, with the ultimate goal of informing the design of more inclusive Voice User Interfaces (VUIs). The authors situate their work within the broader context of rapid advances in speech‑enabled devices (smartphones, smart speakers, in‑car assistants) and note that, despite their growing ubiquity, VUIs still suffer from usability problems such as one‑off interactions, poor discoverability, and difficulty handling noisy or multitasking environments. While mainstream VUIs have begun to be adopted by users with disabilities, research on their accessibility has largely remained within the assistive‑technology domain and has not examined the everyday lived experiences of visually impaired people in depth.

The paper first reviews design frameworks for accessibility (Universal Design, Inclusive Design, Ability‑Based Design) and highlights that most existing guidelines target web content rather than voice‑first interactions. It then discusses the current state of accessible voice interfaces, noting that commercial assistants (Alexa, Siri, Google Assistant) can provide independence for many users but were not designed with accessibility as a primary goal. In the assistive‑technology space, screen readers (JAWS, VoiceOver) and speech‑recognition tools (Dragon) are widely used, yet their effectiveness depends heavily on the underlying application’s compliance with accessibility standards.

Methodologically, the study adopts a classic ethnographic approach combined with conversation‑analytic techniques (EMCA). The researcher immersed themselves in a UK charity that supports blind and partially sighted individuals, gaining access through staff gatekeepers. Initial activities included attending a sight‑loss awareness training, using “simulation spectacles” and blindfolds to experience visual limitations, and observing both informal social gatherings (fortnightly meet‑ups) and structured IT drop‑in sessions (twice weekly). Data were collected via field notes, contextual interviews, and participant observation of technology use (screen readers, voice assistants, audio watches, CD‑based news services).

Early findings reveal considerable heterogeneity among participants. Visual conditions range from cataract‑induced blur to glaucoma‑related peripheral loss, leading to diverse preferences: some blind users rely exclusively on screen readers and voice assistants, while partially sighted users combine enlarged fonts with occasional voice commands. Many participants also have co‑occurring motor or hearing impairments, further shaping their interaction strategies. Communication practices differ as well; blind individuals depend on vocal cues and explicit turn‑taking, whereas partially sighted people still use body language. Practical work‑arounds include accessing less‑cluttered desktop versions of websites via JAWS, shopping at off‑peak hours to avoid crowds, using audio‑enabled watches, and receiving local news on CDs.

Attitudes toward technology vary with age, prior experience, and perceived effort. Some older users feel overwhelmed by learning new devices, especially when visual loss is progressive, while others appreciate the convenience of voice assistants for scheduling or reminders. Staff and volunteers who are themselves visually impaired play a crucial role in providing peer guidance. Financial barriers remain a significant obstacle, as high‑cost assistive hardware (desktop screen readers, specialized phones) limits adoption.

The authors conclude that the ethnographic insights gathered will inform a set of design recommendations for VUIs that accommodate both permanent visual impairments and situationally‑induced challenges. They propose future work that includes developing prototype voice interfaces, testing them in real‑world contexts, and exploring low‑cost, open‑source speech technologies to mitigate financial constraints. Overall, the paper underscores the importance of grounding VUI design in the lived practices of visually impaired users, advocating for interdisciplinary collaboration and policy support to achieve truly inclusive voice‑first experiences.