Intro
Voice user interface (VUI) devices are gaining popularity and have a promising future to open doors for technology to be accessible for everyone. They are made possible through natural language processing that enables the system to understand human language. In this study, we aimed to explore how students, whose first language was not English, interact with their VUI devices.
Most of the literature written about VUI user experience focuses on best practices and challenges for design with predominantly English-language centered audiences. It is difficult to find research on what populations where English is a second language experience when using VUI devices. Because of this,our research question is framed to help us learn more about what these experiences are like for University level ESL students. Our research question was, “What is the experience of voice technology for students for whom English is their second language?”
Process
Survey
We used a screener to recruit participants who fulfilled two different qualifications: (1) they currently use a VUI device within their everyday lives, and (2) English is not their native language. The screener also collected data on the type of VUI devices used in relation to task and environments, as well as current frustrations with VUI products.
Field Studies
We scheduled 45-minute participant sessions with participants selected through our screener and employed two field study methods:
- On-site semi-structured interviews
- Observations
Sessions were at participants homes with a facilitator and note taker present. Data was collected through note taking, video recordings, and participants sketches. Observations were conducted in the room the VUI was primarily used.
The facilitator opened with neutral questions, followed by an activity asking participants to sketch out common experiences with their VUI devices. This helped get participants to think of their common interactions and surrounding context in greater detail to avoid recall bias. The facilitator used a script to guide the interview process, but had flexibility to explore salient topics that built on answers brought up in the moment. Throughout the interviews, video data was collected of participants doing common tasks with their VUI devices.
Analysis
The interview scripts were transcribed and underwent a thematic analysis based on salient moments and keywords. The title of each theme had to be an actual word or phrase used by the participants themselves in order to prevent researcher bias about their meaning and intent.
This data was then categorized into core higher-level themes and identifying sub-themes
Limitations
While we feel our project provided strong insights on VUI interactions with students who speak English as a second language, our research did have some limitations. The most impactful limitation was the diversity and size of our participant sample. Ideally this project would have more time and resources to recruit multiple participants from more than three non-U.S. countries (our subjects’ native languages originate from Thailand, Mexico, and South Korea). With a larger more diverse participant pool our team could potentially see if there is concurrent data from participants who have similar regional language locations. Our data was also limited by the lack of location diversity. All of our observation studies took place in user’s homes, and we did not observe interactions with VUI devices on mobile phones or in automobiles. The research sessions were scheduled based on participant availability, making it impossible to capture many of the VUI tasks discussed during the usual time and setting they would occur. If this was a longer term study an at home diary study and recordings could remedy this and potentially capture more understandings.
Our findings also made conjectures about American cultures based on our own personal bias. It would have been ideal to break down our own perceptions and definitions of American culture and do more preliminary research on the subject before integrating such assumptions into our findings.