English is widely regarded as one of the more difficult languages to learn, but it is one of the most commonly spoken second languages. English language students have access to many resources to practice their reading and writing skills, but very little exists to help people with their speaking.
CrowdSpeak is the solution. Through crowdsourced feedback, users will get feedback on their spoken English.
University of Rochester
February - April 2017
I worked as a developer, coder, and designer. I led group efforts on needfinding and helped create the final prototype.
We created a prototype that served as a crowdsourced-based online auto-transcription tool. This prototype provides individuals a service for feedback by generating a record of what they said, which is then corrected by other users with focus on tone, volume, cadence, word choice, etc.
To aid us with our research and to provide feedback for our eventual product, we conducted surveys of the general University of Rochester student body. In addition, we conducted a small number of more focused personal interviews. We wanted to know what aspects of speech were the most important to people from these options: Body Language, Speed, Tone, Volume, and Word Choice.
Although the majority of our respondents were native English speakers, we still gained a sizeable amount of data and feedback about what people prioritize when using English/language in general. Word choice and speed were the components that people wanted the most feedback about.
While both native and non-native English speakers had similar preferences for body language and volume feedback, non-native English speakers tended to identify feedback for word choice and tone as more important than native English speakers did. Native English speakers identified speed as a more important category than non-native English speakers.
We made three prototypes - a paper prototype, a wireframe, and (finally) the web application.
Paper Prototype - Through the paper prototype, we figured out basic functionality for a web application. With this testing, we made a smooth coordination between viewing a transcription and editing a transcription. We also developed a method for color-coding comments and changes to help language learners. We made two paper-prototypes, one for each type of user: language learners and transcription editors (people who provide the crowdsourced corrections).
Wireframe Prototype - This prototype was a mock-up of our paper prototype so we could see how it would look on a web browser. We had users look at the above image and narrate how they would navigate through the pages.